Failed to convert self trained YOLO11m model

Application Development Algorithm Toolchain
1

I am trying to convert my custom-trained YOLO11m model into a BIN model. I executed the following command, but I was unable to convert it successfully.
root@00879d456b7e:/open_explorer/yolo11m# python3 mapper.py --onnx ./best.onnx --cal-images ../Robomaster.v3i.yolov11/train/images/
On the right is an algorithm for the BIN file downloaded from the model zoo, and the left diagram shows the model after converting my trained model into a BIN format.

rdk-yolo11m-convert
rdk-yolo11m-convert3840×2160 506 KB

root@00879d456b7e:/open_explorer/yolo11m# python3 mapper.py --onnx ./best.onnx --cal-images ../Robomaster.v3i.yolov11/train/images/
-AresVs-_BO2_2_133_jpg.rf.7c4fa8539d8b532e36cbe635d66c760c.jpg
-AresVs-_BO2_2_157_jpg.rf.6d3eeb217750956deb742e5b450b3484.jpg
-AresVs-_BO2_2_197_jpg.rf.1f97563c31dd40e53e3f061d541f1add.jpg
-AresVs-_BO2_2_253_jpg.rf.6e5a239db406eceee1e4866c6fe4bcc5.jpg
-AresVs-_BO2_2_314_jpg.rf.ef8dfe46652efba50733670f8bde4468.jpg
-AresVs-_BO2_2_317_jpg.rf.5a2251c8e0a79630f390cdae322e6a21.jpg
-AresVs-_BO2_2_328_jpg.rf.ae8f14bea160c837a5895d7f64bf5edf.jpg
-AresVs-_BO2_2_375_jpg.rf.6c96782f406e50b5e9ba66cada8f6202.jpg
-AresVs-_BO2_2_386_jpg.rf.6a90888e8cd640b64ce27b02340f0b61.jpg
-AresVs-_BO2_2_56_jpg.rf.5cc429a0efba2ac35376f9af2e474836.jpg
-VsBorn-of-Fire_BO2_1_222_jpg.rf.500384bb0b9e1de0330f5f1b2a307dc6.jpg
-VsBorn-of-Fire_BO2_1_241_jpg.rf.c44de2c74dfb8a85557000e28c5ccca5.jpg
-VsBorn-of-Fire_BO2_1_256_jpg.rf.adc468844e5ebb5a54706a42af1a5d5d.jpg
-VsBorn-of-Fire_BO2_1_310_jpg.rf.5f26b20204ea55cb474fedecfe35e836.jpg
-VsBorn-of-Fire_BO2_1_324_jpg.rf.1a33ae0709f113776344d2b7937f6567.jpg
-VsCUBOT_BO2_1_103_jpg.rf.6bf02c074016c7535971b6a5d5ec7443.jpg
-VsCUBOT_BO2_1_139_jpg.rf.074b080ef43f72e7c4e88a1c3b7022e7.jpg
-VsCUBOT_BO2_1_152_jpg.rf.779bd0805c4913739c6a81c015e1cc22.jpg
-VsCUBOT_BO2_1_155_jpg.rf.9275351f8d57009fdd252802d70a72b1.jpg
-VsCUBOT_BO2_1_20_jpg.rf.1bf81d37b845485b94c7387414344778.jpg
-VsCUBOT_BO2_1_226_jpg.rf.7b772d2352d0cb760f287b26bd1c8cf2.jpg
-VsCUBOT_BO2_1_232_jpg.rf.74014e0c631ecab3ea204133ac817e23.jpg
-VsCUBOT_BO2_1_274_jpg.rf.92578f91f24762e7d3190b6ec7ce343c.jpg
-VsCUBOT_BO2_1_87_jpg.rf.0c9ce638b86343622c4ebaaacb4f5f96.jpg
-VsHLL_BO2_2_144_jpg.rf.9da05d7f6306eb7862828b3f5fcf3c9e.jpg
-VsHLL_BO2_2_214_jpg.rf.4820d2ab1dd9d46570e99142433147c6.jpg
-VsHLL_BO2_2_72_jpg.rf.a37c3cd88bef1be61bccabf32a51cc64.jpg
-VsHLL_BO2_2_73_jpg.rf.c7f2b4ebc03ef1bbe13a791696546b56.jpg
-VsHLL_BO2_2_7_jpg.rf.bc2f5c0fa47695eefd28d1106bd0aa23.jpg
-VsHLL_BO2_2_81_jpg.rf.0be274923ca9bf3d7fecea3bb0f28327.jpg
-VsRPS_BO2_2_103_jpg.rf.b191f497b8ecc180b7248fd40703d09d.jpg
-VsRPS_BO2_2_106_jpg.rf.57a69fddd040025c41f6312a88aaf4e1.jpg
-VsRPS_BO2_2_108_jpg.rf.5040ad4f02c22b29c3c659a748785c6e.jpg
-VsRPS_BO2_2_16_jpg.rf.ef7e0091d197205c5894445a6f6336c2.jpg
-VsRPS_BO2_2_211_jpg.rf.0c2df54b58395784b8a7b9acfb4391b8.jpg
-VsRPS_BO2_2_248_jpg.rf.b79239d445e8d50d0e9d3758915d2269.jpg
AllianceVsArtisans_BO2_2_209_jpg.rf.63281e06019f74334277cb62045dae88.jpg
AllianceVsArtisans_BO2_2_246_jpg.rf.b31fa512da36f554d4753a0fd592cc74.jpg
AllianceVsArtisans_BO2_2_24_jpg.rf.14138885ef58a72e693b44da3fa7834f.jpg
AllianceVsArtisans_BO2_2_251_jpg.rf.c35a56c74e6bb42f6f0c7d7022c384d7.jpg
AllianceVsArtisans_BO2_2_257_jpg.rf.7a313b89dee180b69a6cfff2ffa8555e.jpg
AllianceVsArtisans_BO2_2_260_jpg.rf.9edd2e6652bc231d86194281adf537f8.jpg
root@00879d456b7e:/open_explorer/yolo11m# python3 mapper.py --onnx ./best.onnx --cal-images ../Robomaster.v3i.yolov11/train/images/
[MZOO] [21:25:16.100] [INFO] Namespace(cal_images='../Robomaster.v3i.yolov11/train/images/', onnx='./best.onnx', output_dir='.', quantized='int8', jobs=16, optimize_level='O3', cal_sample=True, cal_sample_num=20, save_cache=False, cal='.calibration_data_temporary_folder', ws='.temporary_workspace')
[MZOO] [21:25:16.100] [INFO] Output directory set to ONNX file directory: /open_explorer/yolo11m
[MZOO] [21:25:16.100] [INFO] Resolved paths:
[MZOO] [21:25:16.100] [INFO]   ONNX model: /open_explorer/yolo11m/best.onnx
[MZOO] [21:25:16.100] [INFO]   Calibration images: /open_explorer/Robomaster.v3i.yolov11/train/images
[MZOO] [21:25:16.100] [INFO]   Output directory: /open_explorer/yolo11m
[MZOO] [21:25:16.100] [INFO]   Workspace: /open_explorer/yolo11m/.temporary_workspace
[MZOO] [21:25:18.208] [INFO] hb_mapper is available and working.
[MZOO] [21:25:18.209] [INFO] Loading ONNX model from: /open_explorer/yolo11m/best.onnx
[MZOO] [21:25:18.518] [DEBUG] Model has a single input.
[MZOO] [21:25:18.518] [DEBUG] Input data type is float32 (tensor(float)).
[MZOO] [21:25:18.518] [DEBUG] NCHW check success.
[MZOO] [21:25:18.518] [DEBUG] ONNX Runtime session released.
[MZOO] [21:25:18.525] [INFO] Sampling enabled. Sampled 20 images from 50 total images.
[MZOO] [21:25:18.526] [INFO] Workspace '/open_explorer/yolo11m/.temporary_workspace' created successfully.
[MZOO] [21:25:18.526] [INFO] Configuration file created: /open_explorer/yolo11m/.temporary_workspace/config.yaml
[MZOO] [21:25:18.526] [INFO] Preparing calibration data...
[MZOO] [21:25:19.017] [INFO] Calibration data has been successfully generated.
[MZOO] [21:25:19.017] [INFO] Changed working directory to: /open_explorer/yolo11m/.temporary_workspace
[MZOO] [21:25:19.017] [INFO] Executing model conversion...
[MZOO] [21:25:19.017] [INFO] Command: hb_mapper makertbin --config config.yaml --model-type onnx
2025-08-30 21:25:20,720 INFO log will be stored in /open_explorer/yolo11m/.temporary_workspace/hb_mapper_makertbin.log
2025-08-30 21:25:20,720 INFO Start hb_mapper....
2025-08-30 21:25:20,720 INFO hbdk version 3.49.15
2025-08-30 21:25:20,721 INFO horizon_nn version 1.1.0
2025-08-30 21:25:20,721 INFO hb_mapper version 1.24.3
2025-08-30 21:25:20,721 INFO Start Model Convert....
2025-08-30 21:25:20,725 INFO Using onnx model file: /open_explorer/yolo11m/best.onnx
2025-08-30 21:25:20,824 INFO Model has 1 inputs according to model file
2025-08-30 21:25:20,824 INFO working_dir does not exist. Creating working_dir: /open_explorer/yolo11m/.temporary_workspace/bpu_model_output
2025-08-30 21:25:20,825 INFO Model name not given in yaml_file, using model name from model file: ['images']
2025-08-30 21:25:20,825 INFO Model input shape not given in yaml_file, using shape from model file: [[1, 3, 640, 640]]
2025-08-30 21:25:20,825 INFO nv12 input type rt received.
2025-08-30 21:25:20,825 WARNING The calibration dir name suffix is not the same as the value float32 of the parameter cal_data_type, the parameter setting will prevail
2025-08-30 21:25:20,825 INFO custom_op does not exist, skipped
2025-08-30 21:25:20,825 WARNING Input node images's input_source not set, it will be set to pyramid by default
2025-08-30 21:25:20,828 INFO *******************************************
2025-08-30 21:25:20,828 INFO First calibration picture name: -AresVs-_BO2_2_157_jpg.rf.6d3eeb217750956deb742e5b450b3484.jpg.rgbchw
2025-08-30 21:25:20,828 INFO First calibration picture md5:
b79f9ee2e06899019244cc97e2ad1ad2  /open_explorer/yolo11m/.temporary_workspace/.calibration_data_temporary_folder/-AresVs-_BO2_2_157_jpg.rf.6d3eeb217750956deb742e5b450b3484.jpg.rgbchw
2025-08-30 21:25:20,861 INFO *******************************************
2025-08-30 21:25:22,323 INFO Start to Horizon NN Model Convert.
2025-08-30 21:25:22,434 INFO Loading horizon_nn debug methods:set()
2025-08-30 21:25:22,435 INFO The activation calibration parameters:
    calibration_type:     ['max', 'kl']
    max_percentile:       [0.99995, 1.0]
    per_channel:          [True, False]
    asymmetric:           [True, False]
The modelwise search parameters:
    similarity:           0.995
    metric:               cosine-similarity
The input of all Softmax nodes are set to : int8
2025-08-30 21:25:22,435 INFO input images is from pyramid. Its layout is set to NHWC
2025-08-30 21:25:22,435 INFO The specified model compilation architecture: bayes-e.
2025-08-30 21:25:22,435 INFO The specified model compilation optimization parameters: [].
2025-08-30 21:25:22,435 INFO Start to prepare the onnx model.
2025-08-30 21:25:22,527 INFO Input ONNX Model Information:
ONNX IR version:          6
Opset version:            ['ai.onnx v11', 'horizon v1']
Producer:                 pytorch v2.8.0
Domain:                   None
Version:                  None
Graph input:
    images:               shape=[1, 3, 640, 640], dtype=FLOAT32
Graph output:
    output0:              shape=[1, 9, 8400], dtype=FLOAT32
2025-08-30 21:25:23,622 INFO End to prepare the onnx model.
2025-08-30 21:25:23,758 INFO Saving model to: best_bayese_640x640_nv12_original_float_model.onnx.
2025-08-30 21:25:23,758 INFO Start to optimize the onnx model.
2025-08-30 21:25:25,087 INFO End to optimize the onnx model.
2025-08-30 21:25:25,215 INFO Saving model to: best_bayese_640x640_nv12_optimized_float_model.onnx.
2025-08-30 21:25:25,215 INFO Start to calibrate the model.
Layer /model.23/dfl/Softmax_reducesum_FROM_QUANTIZED_SOFTMAX
	/model.23/dfl/Softmax_reducesum_FROM_QUANTIZED_SOFTMAX layer, feature size in axis 3 should in range [1, 4096]. But given size 8400
	/model.23/dfl/Softmax_reducesum_FROM_QUANTIZED_SOFTMAX layer, feature size in axis 3 should in range [1, 4096]. But given size 8400

2025-08-30 21:25:25,698 INFO There are 20 samples in the data set.
2025-08-30 21:25:25,699 INFO Run calibration model with modelwise search method.
2025-08-30 21:25:26,557 INFO Calibration using batch 8
calibration in progress:   0%|                                                                                                                                          | 0/3 [00:00<?, ?it/s]2025-08-30 21:25:41.332057349 [E:onnxruntime:, sequential_executor.cc:183 Execute] Non-zero status code returned while running Reshape node. Name:'/model.10/m/m.0/attn/Reshape' Status Message: /home/jenkins/agent/workspace/model_convert_hmct1_ut_cpu_1.1.0/onnxruntime/onnxruntime/core/providers/cpu/tensor/reshape_helper.h:43 onnxruntime::ReshapeHelper::ReshapeHelper(const onnxruntime::TensorShape&, std::vector<long int>&) gsl::narrow_cast<int64_t>(input_shape.Size()) == size was false. The input tensor cannot be reshaped to the requested shape. Input shape:{8,512,20,20}, requested shape:{1,4,128,400}

calibration in progress:   0%|                                                                                                                                          | 0/3 [00:14<?, ?it/s]
2025-08-30 21:25:41,333 INFO Reset batch_size=1 and execute forward again...
calibration in progress: 100%|████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 20/20 [01:49<00:00,  5.50s/it]
Layer /model.23/dfl/Softmax_reducesum_FROM_QUANTIZED_SOFTMAX
	/model.23/dfl/Softmax_reducesum_FROM_QUANTIZED_SOFTMAX layer, feature size in axis 3 should in range [1, 4096]. But given size 8400
	/model.23/dfl/Softmax_reducesum_FROM_QUANTIZED_SOFTMAX layer, feature size in axis 3 should in range [1, 4096]. But given size 8400

Layer /model.23/dfl/Softmax_reducesum_FROM_QUANTIZED_SOFTMAX
	/model.23/dfl/Softmax_reducesum_FROM_QUANTIZED_SOFTMAX layer, feature size in axis 3 should in range [1, 4096]. But given size 8400
	/model.23/dfl/Softmax_reducesum_FROM_QUANTIZED_SOFTMAX layer, feature size in axis 3 should in range [1, 4096]. But given size 8400

Layer /model.23/dfl/Softmax_reducesum_FROM_QUANTIZED_SOFTMAX
	/model.23/dfl/Softmax_reducesum_FROM_QUANTIZED_SOFTMAX layer, feature size in axis 3 should in range [1, 4096]. But given size 8400
	/model.23/dfl/Softmax_reducesum_FROM_QUANTIZED_SOFTMAX layer, feature size in axis 3 should in range [1, 4096]. But given size 8400

Layer /model.23/dfl/Softmax_reducesum_FROM_QUANTIZED_SOFTMAX
	/model.23/dfl/Softmax_reducesum_FROM_QUANTIZED_SOFTMAX layer, feature size in axis 3 should in range [1, 4096]. But given size 8400
	/model.23/dfl/Softmax_reducesum_FROM_QUANTIZED_SOFTMAX layer, feature size in axis 3 should in range [1, 4096]. But given size 8400

Layer /model.23/dfl/Softmax_reducesum_FROM_QUANTIZED_SOFTMAX
	/model.23/dfl/Softmax_reducesum_FROM_QUANTIZED_SOFTMAX layer, feature size in axis 3 should in range [1, 4096]. But given size 8400
	/model.23/dfl/Softmax_reducesum_FROM_QUANTIZED_SOFTMAX layer, feature size in axis 3 should in range [1, 4096]. But given size 8400

Layer /model.23/dfl/Softmax_reducesum_FROM_QUANTIZED_SOFTMAX
	/model.23/dfl/Softmax_reducesum_FROM_QUANTIZED_SOFTMAX layer, feature size in axis 3 should in range [1, 4096]. But given size 8400
	/model.23/dfl/Softmax_reducesum_FROM_QUANTIZED_SOFTMAX layer, feature size in axis 3 should in range [1, 4096]. But given size 8400

2025-08-30 21:28:04,840 INFO Select max-percentile:percentile=0.99995 method.
Layer /model.23/dfl/Softmax_reducesum_FROM_QUANTIZED_SOFTMAX
	/model.23/dfl/Softmax_reducesum_FROM_QUANTIZED_SOFTMAX layer, feature size in axis 3 should in range [1, 4096]. But given size 8400
	/model.23/dfl/Softmax_reducesum_FROM_QUANTIZED_SOFTMAX layer, feature size in axis 3 should in range [1, 4096]. But given size 8400

2025-08-30 21:28:07,056 INFO End to calibrate the model.
2025-08-30 21:28:07,353 INFO Saving model to: best_bayese_640x640_nv12_calibrated_model.onnx.
2025-08-30 21:28:07,353 INFO Start to quantize the model.
Layer /model.23/dfl/Softmax_reducesum_FROM_QUANTIZED_SOFTMAX
	/model.23/dfl/Softmax_reducesum_FROM_QUANTIZED_SOFTMAX layer, feature size in axis 3 should in range [1, 4096]. But given size 8400
	/model.23/dfl/Softmax_reducesum_FROM_QUANTIZED_SOFTMAX layer, feature size in axis 3 should in range [1, 4096]. But given size 8400

Layer /model.23/dfl/Softmax_reducesum_FROM_QUANTIZED_SOFTMAX
	/model.23/dfl/Softmax_reducesum_FROM_QUANTIZED_SOFTMAX layer, feature size in axis 3 should in range [1, 4096]. But given size 8400
	/model.23/dfl/Softmax_reducesum_FROM_QUANTIZED_SOFTMAX layer, feature size in axis 3 should in range [1, 4096]. But given size 8400

Layer /model.23/dfl/Softmax_reducesum_FROM_QUANTIZED_SOFTMAX
	/model.23/dfl/Softmax_reducesum_FROM_QUANTIZED_SOFTMAX layer, feature size in axis 3 should in range [1, 4096]. But given size 8400
	/model.23/dfl/Softmax_reducesum_FROM_QUANTIZED_SOFTMAX layer, feature size in axis 3 should in range [1, 4096]. But given size 8400

2025-08-30 21:28:13,381 INFO End to quantize the model.
2025-08-30 21:28:13,499 INFO Saving model to: best_bayese_640x640_nv12_quantized_model.onnx.
2025-08-30 21:28:13,499 INFO Start to compile the model with march bayes-e.
2025-08-30 21:28:14,110 INFO Compile submodel: main_graph_subgraph_0
2025-08-30 21:28:14,198 INFO hbdk-cc parameters:['--O3', '--debug', '--core-num', '1', '--fast', '--jobs', '16', '--input-layout', 'NHWC', '--output-layout', 'NHWC', '--input-source', 'pyramid']
[==================================================] 100%
2025-08-30 21:38:14,747 INFO consumed time 600.517
2025-08-30 21:38:15,162 INFO FPS=28.21, latency = 35445.0 us, DDR = 137015296 bytes   (see main_graph_subgraph_0.html)
2025-08-30 21:38:15,208 INFO Compile submodel: main_graph_subgraph_1
2025-08-30 21:38:15,228 INFO hbdk-cc parameters:['--O3', '--debug', '--core-num', '1', '--fast', '--jobs', '16', '--input-layout', 'NHWC', '--output-layout', 'NHWC', '--input-source', 'ddr,ddr,ddr']
2025-08-30 21:39:36,231 INFO consumed time 80.9697
2025-08-30 21:39:36,307 INFO FPS=315.27, latency = 3171.9 us, DDR = 17490592 bytes   (see main_graph_subgraph_1.html)
2025-08-30 21:39:36,977 INFO End to compile the model with march bayes-e.
2025-08-30 21:39:47,914 INFO The main quantized node information:
======================================================================================================================================
Node                                                ON   Subgraph  Type                       Cosine Similarity  Threshold  DataType  
--------------------------------------------------------------------------------------------------------------------------------------
HZ_PREPROCESS_FOR_images                            BPU  id(0)     HzSQuantizedPreprocess     0.999957           127.0      int8      
/model.0/conv/Conv                                  BPU  id(0)     HzSQuantizedConv           0.996449           1.03386    int8      
/model.0/act/Mul                                    BPU  id(0)     HzLut                      0.996207           55.1878    int8      
/model.1/conv/Conv                                  BPU  id(0)     HzSQuantizedConv           0.995699           49.2212    int8      
/model.1/act/Mul                                    BPU  id(0)     HzLut                      0.993340           49.531     int8      
/model.2/cv1/conv/Conv                              BPU  id(0)     HzSQuantizedConv           0.992912           34.2121    int8      
/model.2/cv1/act/Mul                                BPU  id(0)     HzLut                      0.990032           33.6346    int8      
/model.2/Split                                      BPU  id(0)     Split                      0.987279           11.9327    int8      
/model.2/m.0/cv1/conv/Conv                          BPU  id(0)     HzSQuantizedConv           0.996169           11.9327    int8      
/model.2/m.0/cv2/conv/Conv                          BPU  id(0)     HzSQuantizedConv           0.991514           11.9327    int8      
/model.2/m.0/cv1/act/Mul                            BPU  id(0)     HzLut                      0.995338           14.945     int8      
/model.2/m.0/cv2/act/Mul                            BPU  id(0)     HzLut                      0.972692           18.8007    int8      
/model.2/m.0/m/m.0/cv1/conv/Conv                    BPU  id(0)     HzSQuantizedConv           0.989375           5.57777    int8      
/model.2/m.0/m/m.0/cv1/act/Mul                      BPU  id(0)     HzLut                      0.988867           12.747     int8      
/model.2/m.0/m/m.0/cv2/conv/Conv                    BPU  id(0)     HzSQuantizedConv           0.993607           9.18352    int8      
/model.2/m.0/m/m.0/cv2/act/Mul                      BPU  id(0)     HzLut                      0.995112           10.0757    int8      
/model.2/m.0/m/m.0/Add                              BPU  id(0)     HzSElementwiseAdd          0.996062           5.57777    int8      
/model.2/m.0/m/m.1/cv1/conv/Conv                    BPU  id(0)     HzSQuantizedConv           0.991749           6.29686    int8      
/model.2/m.0/m/m.1/cv1/act/Mul                      BPU  id(0)     HzLut                      0.991948           7.18793    int8      
/model.2/m.0/m/m.1/cv2/conv/Conv                    BPU  id(0)     HzSQuantizedConv           0.993907           4.71358    int8      
/model.2/m.0/m/m.1/cv2/act/Mul                      BPU  id(0)     HzLut                      0.993791           6.66543    int8      
/model.2/m.0/m/m.1/Add                              BPU  id(0)     HzSElementwiseAdd          0.996854           6.29686    int8      
/model.2/m.0/Concat                                 BPU  id(0)     Concat                     0.995867           9.43722    int8      
/model.2/m.0/cv3/conv/Conv                          BPU  id(0)     HzSQuantizedConv           0.981414           9.43722    int8      
/model.2/m.0/cv3/act/Mul                            BPU  id(0)     HzLut                      0.981903           11.7931    int8      
/model.2/Split_output_0_calibrated_Requantize       BPU  id(0)     HzRequantize               --                 --         int8      
/model.2/Split_output_1_calibrated_Requantize       BPU  id(0)     HzRequantize               --                 --         int8      
/model.2/Concat                                     BPU  id(0)     Concat                     0.986495           11.9327    int8      
/model.2/cv2/conv/Conv                              BPU  id(0)     HzSQuantizedConv           0.990360           11.4174    int8      
/model.2/cv2/act/Mul                                BPU  id(0)     HzLut                      0.976560           10.9082    int8      
/model.3/conv/Conv                                  BPU  id(0)     HzSQuantizedConv           0.990717           4.59709    int8      
/model.3/act/Mul                                    BPU  id(0)     HzLut                      0.990748           6.9388     int8      
/model.4/cv1/conv/Conv                              BPU  id(0)     HzSQuantizedConv           0.990868           4.87618    int8      
/model.4/cv1/act/Mul                                BPU  id(0)     HzLut                      0.990278           6.38105    int8      
/model.4/Split                                      BPU  id(0)     Split                      0.991913           4.53303    int8      
/model.4/m.0/cv1/conv/Conv                          BPU  id(0)     HzSQuantizedConv           0.992734           4.53303    int8      
/model.4/m.0/cv2/conv/Conv                          BPU  id(0)     HzSQuantizedConv           0.989221           4.53303    int8      
/model.4/m.0/cv1/act/Mul                            BPU  id(0)     HzLut                      0.992602           4.39424    int8      
/model.4/m.0/cv2/act/Mul                            BPU  id(0)     HzLut                      0.977659           8.25119    int8      
/model.4/m.0/m/m.0/cv1/conv/Conv                    BPU  id(0)     HzSQuantizedConv           0.992381           3.37693    int8      
/model.4/m.0/m/m.0/cv1/act/Mul                      BPU  id(0)     HzLut                      0.984154           6.31164    int8      
/model.4/m.0/m/m.0/cv2/conv/Conv                    BPU  id(0)     HzSQuantizedConv           0.991696           5.01558    int8      
/model.4/m.0/m/m.0/cv2/act/Mul                      BPU  id(0)     HzLut                      0.989659           5.38019    int8      
/model.4/m.0/m/m.0/Add                              BPU  id(0)     HzSElementwiseAdd          0.992404           3.37693    int8      
/model.4/m.0/m/m.1/cv1/conv/Conv                    BPU  id(0)     HzSQuantizedConv           0.992603           4.29768    int8      
/model.4/m.0/m/m.1/cv1/act/Mul                      BPU  id(0)     HzLut                      0.985472           5.63247    int8      
/model.4/m.0/m/m.1/cv2/conv/Conv                    BPU  id(0)     HzSQuantizedConv           0.994739           4.285      int8      
/model.4/m.0/m/m.1/cv2/act/Mul                      BPU  id(0)     HzLut                      0.994640           7.07319    int8      
/model.4/m.0/m/m.1/Add                              BPU  id(0)     HzSElementwiseAdd          0.995297           4.29768    int8      
/model.4/m.0/Concat                                 BPU  id(0)     Concat                     0.994238           7.4582     int8      
/model.4/m.0/cv3/conv/Conv                          BPU  id(0)     HzSQuantizedConv           0.988343           7.4582     int8      
/model.4/m.0/cv3/act/Mul                            BPU  id(0)     HzLut                      0.983935           6.80657    int8      
/model.4/Split_output_0_calibrated_Requantize       BPU  id(0)     HzRequantize               --                 --         int8      
/model.4/Split_output_1_calibrated_Requantize       BPU  id(0)     HzRequantize               --                 --         int8      
/model.4/Concat                                     BPU  id(0)     Concat                     0.988228           4.53303    int8      
/model.4/cv2/conv/Conv                              BPU  id(0)     HzSQuantizedConv           0.992513           4.75003    int8      
/model.4/cv2/act/Mul                                BPU  id(0)     HzLut                      0.982728           6.51416    int8      
/model.5/conv/Conv                                  BPU  id(0)     HzSQuantizedConv           0.993028           4.04468    int8      
/model.5/act/Mul                                    BPU  id(0)     HzLut                      0.985407           6.08988    int8      
/model.6/cv1/conv/Conv                              BPU  id(0)     HzSQuantizedConv           0.991989           4.69882    int8      
/model.6/cv1/act/Mul                                BPU  id(0)     HzLut                      0.983613           6.97167    int8      
/model.6/Split                                      BPU  id(0)     Split                      0.982463           5.42047    int8      
/model.6/m.0/cv1/conv/Conv                          BPU  id(0)     HzSQuantizedConv           0.992454           5.42047    int8      
/model.6/m.0/cv2/conv/Conv                          BPU  id(0)     HzSQuantizedConv           0.988931           5.42047    int8      
/model.6/m.0/cv1/act/Mul                            BPU  id(0)     HzLut                      0.991571           5.13887    int8      
/model.6/m.0/cv2/act/Mul                            BPU  id(0)     HzLut                      0.981411           8.44725    int8      
/model.6/m.0/m/m.0/cv1/conv/Conv                    BPU  id(0)     HzSQuantizedConv           0.994792           4.5099     int8      
/model.6/m.0/m/m.0/cv1/act/Mul                      BPU  id(0)     HzLut                      0.985632           7.00239    int8      
/model.6/m.0/m/m.0/cv2/conv/Conv                    BPU  id(0)     HzSQuantizedConv           0.993591           5.76958    int8      
/model.6/m.0/m/m.0/cv2/act/Mul                      BPU  id(0)     HzLut                      0.987164           6.59678    int8      
/model.6/m.0/m/m.0/Add                              BPU  id(0)     HzSElementwiseAdd          0.988994           4.5099     int8      
/model.6/m.0/m/m.1/cv1/conv/Conv                    BPU  id(0)     HzSQuantizedConv           0.994612           5.22575    int8      
/model.6/m.0/m/m.1/cv1/act/Mul                      BPU  id(0)     HzLut                      0.986061           6.57288    int8      
/model.6/m.0/m/m.1/cv2/conv/Conv                    BPU  id(0)     HzSQuantizedConv           0.990659           5.40215    int8      
/model.6/m.0/m/m.1/cv2/act/Mul                      BPU  id(0)     HzLut                      0.990359           10.2801    int8      
/model.6/m.0/m/m.1/Add                              BPU  id(0)     HzSElementwiseAdd          0.990510           5.22575    int8      
/model.6/m.0/Concat                                 BPU  id(0)     Concat                     0.989353           9.33864    int8      
/model.6/m.0/cv3/conv/Conv                          BPU  id(0)     HzSQuantizedConv           0.992059           9.33864    int8      
/model.6/m.0/cv3/act/Mul                            BPU  id(0)     HzLut                      0.982843           7.47056    int8      
/model.6/Split_output_0_calibrated_Requantize       BPU  id(0)     HzRequantize               --                 --         int8      
/model.6/Split_output_1_calibrated_Requantize       BPU  id(0)     HzRequantize               --                 --         int8      
/model.6/Concat                                     BPU  id(0)     Concat                     0.983251           5.42047    int8      
/model.6/cv2/conv/Conv                              BPU  id(0)     HzSQuantizedConv           0.992748           5.53007    int8      
/model.6/cv2/act/Mul                                BPU  id(0)     HzLut                      0.982641           7.72004    int8      
/model.7/conv/Conv                                  BPU  id(0)     HzSQuantizedConv           0.993502           5.9629     int8      
/model.7/act/Mul                                    BPU  id(0)     HzLut                      0.988316           7.66711    int8      
/model.8/cv1/conv/Conv                              BPU  id(0)     HzSQuantizedConv           0.991408           7.5765     int8      
/model.8/cv1/act/Mul                                BPU  id(0)     HzLut                      0.988882           10.8075    int8      
/model.8/Split                                      BPU  id(0)     Split                      0.989294           9.67915    int8      
/model.8/m.0/cv1/conv/Conv                          BPU  id(0)     HzSQuantizedConv           0.993276           9.67915    int8      
/model.8/m.0/cv2/conv/Conv                          BPU  id(0)     HzSQuantizedConv           0.991368           9.67915    int8      
/model.8/m.0/cv1/act/Mul                            BPU  id(0)     HzLut                      0.991976           9.41933    int8      
/model.8/m.0/cv2/act/Mul                            BPU  id(0)     HzLut                      0.983695           9.54043    int8      
/model.8/m.0/m/m.0/cv1/conv/Conv                    BPU  id(0)     HzSQuantizedConv           0.993851           6.53687    int8      
/model.8/m.0/m/m.0/cv1/act/Mul                      BPU  id(0)     HzLut                      0.990950           8.67907    int8      
/model.8/m.0/m/m.0/cv2/conv/Conv                    BPU  id(0)     HzSQuantizedConv           0.994858           7.16897    int8      
/model.8/m.0/m/m.0/cv2/act/Mul                      BPU  id(0)     HzLut                      0.992554           7.23701    int8      
/model.8/m.0/m/m.0/Add                              BPU  id(0)     HzSElementwiseAdd          0.992623           6.53687    int8      
/model.8/m.0/m/m.1/cv1/conv/Conv                    BPU  id(0)     HzSQuantizedConv           0.995993           7.9071     int8      
/model.8/m.0/m/m.1/cv1/act/Mul                      BPU  id(0)     HzLut                      0.993752           6.57143    int8      
/model.8/m.0/m/m.1/cv2/conv/Conv                    BPU  id(0)     HzSQuantizedConv           0.996843           6.32607    int8      
/model.8/m.0/m/m.1/cv2/act/Mul                      BPU  id(0)     HzLut                      0.996098           9.73752    int8      
/model.8/m.0/m/m.1/Add                              BPU  id(0)     HzSElementwiseAdd          0.994680           7.9071     int8      
/model.8/m.0/Concat                                 BPU  id(0)     Concat                     0.994023           12.9702    int8      
/model.8/m.0/cv3/conv/Conv                          BPU  id(0)     HzSQuantizedConv           0.994979           12.9702    int8      
/model.8/m.0/cv3/act/Mul                            BPU  id(0)     HzLut                      0.993481           7.0884     int8      
/model.8/Split_output_0_calibrated_Requantize       BPU  id(0)     HzRequantize               --                 --         int8      
/model.8/Split_output_1_calibrated_Requantize       BPU  id(0)     HzRequantize               --                 --         int8      
/model.8/Concat                                     BPU  id(0)     Concat                     0.990362           9.67915    int8      
/model.8/cv2/conv/Conv                              BPU  id(0)     HzSQuantizedConv           0.992897           9.26783    int8      
/model.8/cv2/act/Mul                                BPU  id(0)     HzLut                      0.991604           9.42389    int8      
/model.9/cv1/conv/Conv                              BPU  id(0)     HzSQuantizedConv           0.997110           8.3216     int8      
/model.9/cv1/act/Mul                                BPU  id(0)     HzLut                      0.996466           8.25414    int8      
/model.9/m/MaxPool                                  BPU  id(0)     HzQuantizedMaxPool         0.998387           10.7857    int8      
/model.9/m_1/MaxPool                                BPU  id(0)     HzQuantizedMaxPool         0.998623           10.7857    int8      
/model.9/m_2/MaxPool                                BPU  id(0)     HzQuantizedMaxPool         0.998677           10.7857    int8      
/model.9/Concat                                     BPU  id(0)     Concat                     0.998446           10.7857    int8      
/model.9/cv2/conv/Conv                              BPU  id(0)     HzSQuantizedConv           0.997253           10.7857    int8      
/model.9/cv2/act/Mul                                BPU  id(0)     HzLut                      0.994220           6.62893    int8      
/model.10/cv1/conv/Conv                             BPU  id(0)     HzSQuantizedConv           0.994713           4.34881    int8      
/model.10/cv1/act/Mul                               BPU  id(0)     HzLut                      0.994447           6.38606    int8      
/model.10/Split                                     BPU  id(0)     Split                      0.993615           5.84506    int8      
/model.10/m/m.0/attn/qkv/conv/Conv                  BPU  id(0)     HzSQuantizedConv           0.995927           5.84506    int8      
/model.10/m/m.0/attn/Reshape                        BPU  id(0)     Reshape                    0.995927           5.56271    int8      
/model.10/m/m.0/attn/Split                          BPU  id(0)     Split                      0.996516           5.56271    int8      
/model.10/m/m.0/attn/Transpose                      BPU  id(0)     Transpose                  0.996516           5.56271    int8      
/model.10/m/m.0/attn/Reshape_2                      BPU  id(0)     Reshape                    0.995585           5.56271    int8      
/model.10/m/m.0/attn/MatMul                         BPU  id(0)     HzSQuantizedMatmul         0.995309           5.56271    int8      
/model.10/m/m.0/attn/Mul                            BPU  id(0)     HzSQuantizedConv           0.995309           92.7044    int8      
...0/attn/Softmax_reducemax_FROM_QUANTIZED_SOFTMAX  BPU  id(0)     HzQuantizedReduceMax       0.998738           16.388     int8      
...0/m/m.0/attn/Softmax_sub_FROM_QUANTIZED_SOFTMAX  BPU  id(0)     HzSElementwiseSub          0.932335           16.388     int8      
...0/m/m.0/attn/Softmax_exp_FROM_QUANTIZED_SOFTMAX  BPU  id(0)     HzLut                      0.985607           5.64418    int8      
...0/attn/Softmax_reducesum_FROM_QUANTIZED_SOFTMAX  BPU  id(0)     HzSQuantizedReduceSum      0.991668           1.0        int8      
.../attn/Softmax_reciprocal_FROM_QUANTIZED_SOFTMAX  BPU  id(0)     HzLut                      0.986513           346.034    int8      
...0/m/m.0/attn/Softmax_mul_FROM_QUANTIZED_SOFTMAX  BPU  id(0)     HzSElementwiseMul          0.978361           1.0        int8      
/model.10/m/m.0/attn/Transpose_1                    BPU  id(0)     Transpose                  0.978363           0.155275   int8      
/model.10/m/m.0/attn/MatMul_1                       BPU  id(0)     HzSQuantizedMatmul         0.987081           5.56271    int8      
/model.10/m/m.0/attn/Reshape_1                      BPU  id(0)     Reshape                    0.987081           4.93744    int8      
/model.10/m/m.0/attn/pe/conv/Conv                   BPU  id(0)     HzSQuantizedConv           0.995635           5.56271    int8      
/model.10/m/m.0/attn/proj/conv/Conv                 BPU  id(0)     HzSQuantizedConv           0.980588           4.10945    int8      
/model.10/m/m.0/ffn/ffn.0/conv/Conv                 BPU  id(0)     HzSQuantizedConv           0.988607           6.89125    int8      
/model.10/m/m.0/ffn/ffn.0/act/Mul                   BPU  id(0)     HzLut                      0.972391           5.21373    int8      
/model.10/m/m.0/ffn/ffn.1/conv/Conv                 BPU  id(0)     HzSQuantizedConv           0.991743           2.98986    int8      
/model.10/Split_output_0_calibrated_Requantize      BPU  id(0)     HzRequantize               --                 --         int8      
/model.10/Concat                                    BPU  id(0)     Concat                     0.988110           5.84506    int8      
/model.10/cv2/conv/Conv                             BPU  id(0)     HzSQuantizedConv           0.994035           6.41853    int8      
/model.10/cv2/act/Mul                               BPU  id(0)     HzLut                      0.987125           4.67183    int8      
/model.11/Resize                                    BPU  id(0)     HzQuantizedResizeUpsample  0.987150           5.68315    int8      
/model.11/Resize_output_0_calibrated_Requantize     BPU  id(0)     HzRequantize               --                 --         int8      
...el.6/cv2/act/Mul_output_0_calibrated_Requantize  BPU  id(0)     HzRequantize               --                 --         int8      
/model.12/Concat                                    BPU  id(0)     Concat                     0.983760           5.68315    int8      
/model.13/cv1/conv/Conv                             BPU  id(0)     HzSQuantizedConv           0.993191           5.33763    int8      
/model.13/cv1/act/Mul                               BPU  id(0)     HzLut                      0.987194           6.77288    int8      
/model.13/Split                                     BPU  id(0)     Split                      0.986247           4.99407    int8      
/model.13/m.0/cv1/conv/Conv                         BPU  id(0)     HzSQuantizedConv           0.993420           4.99407    int8      
/model.13/m.0/cv2/conv/Conv                         BPU  id(0)     HzSQuantizedConv           0.984568           4.99407    int8      
/model.13/m.0/cv1/act/Mul                           BPU  id(0)     HzLut                      0.991360           5.20396    int8      
/model.13/m.0/cv2/act/Mul                           BPU  id(0)     HzLut                      0.979115           9.81083    int8      
/model.13/m.0/m/m.0/cv1/conv/Conv                   BPU  id(0)     HzSQuantizedConv           0.993051           3.25027    int8      
/model.13/m.0/m/m.0/cv1/act/Mul                     BPU  id(0)     HzLut                      0.985830           6.38305    int8      
/model.13/m.0/m/m.0/cv2/conv/Conv                   BPU  id(0)     HzSQuantizedConv           0.990652           4.60144    int8      
/model.13/m.0/m/m.0/cv2/act/Mul                     BPU  id(0)     HzLut                      0.988940           7.62279    int8      
/model.13/m.0/m/m.0/Add                             BPU  id(0)     HzSElementwiseAdd          0.989154           3.25027    int8      
/model.13/m.0/m/m.1/cv1/conv/Conv                   BPU  id(0)     HzSQuantizedConv           0.993720           6.79428    int8      
/model.13/m.0/m/m.1/cv1/act/Mul                     BPU  id(0)     HzLut                      0.985438           6.2065     int8      
/model.13/m.0/m/m.1/cv2/conv/Conv                   BPU  id(0)     HzSQuantizedConv           0.989897           4.26048    int8      
/model.13/m.0/m/m.1/cv2/act/Mul                     BPU  id(0)     HzLut                      0.990443           6.11869    int8      
/model.13/m.0/m/m.1/Add                             BPU  id(0)     HzSElementwiseAdd          0.990890           6.79428    int8      
/model.13/m.0/Concat                                BPU  id(0)     Concat                     0.989296           8.76555    int8      
/model.13/m.0/cv3/conv/Conv                         BPU  id(0)     HzSQuantizedConv           0.991173           8.76555    int8      
/model.13/m.0/cv3/act/Mul                           BPU  id(0)     HzLut                      0.983571           5.55189    int8      
/model.13/Split_output_0_calibrated_Requantize      BPU  id(0)     HzRequantize               --                 --         int8      
/model.13/Split_output_1_calibrated_Requantize      BPU  id(0)     HzRequantize               --                 --         int8      
/model.13/Concat                                    BPU  id(0)     Concat                     0.985904           4.99407    int8      
/model.13/cv2/conv/Conv                             BPU  id(0)     HzSQuantizedConv           0.994381           4.72155    int8      
/model.13/cv2/act/Mul                               BPU  id(0)     HzLut                      0.988244           6.65378    int8      
/model.14/Resize                                    BPU  id(0)     HzQuantizedResizeUpsample  0.988179           5.36634    int8      
/model.14/Resize_output_0_calibrated_Requantize     BPU  id(0)     HzRequantize               --                 --         int8      
...el.4/cv2/act/Mul_output_0_calibrated_Requantize  BPU  id(0)     HzRequantize               --                 --         int8      
/model.15/Concat                                    BPU  id(0)     Concat                     0.983964           5.36634    int8      
/model.16/cv1/conv/Conv                             BPU  id(0)     HzSQuantizedConv           0.994656           3.87582    int8      
/model.16/cv1/act/Mul                               BPU  id(0)     HzLut                      0.994164           4.73331    int8      
/model.16/Split                                     BPU  id(0)     Split                      0.994110           2.70114    int8      
/model.16/m.0/cv1/conv/Conv                         BPU  id(0)     HzSQuantizedConv           0.994845           2.70114    int8      
/model.16/m.0/cv2/conv/Conv                         BPU  id(0)     HzSQuantizedConv           0.981503           2.70114    int8      
/model.16/m.0/cv1/act/Mul                           BPU  id(0)     HzLut                      0.995319           3.27481    int8      
/model.16/m.0/cv2/act/Mul                           BPU  id(0)     HzLut                      0.977296           9.56754    int8      
/model.16/m.0/m/m.0/cv1/conv/Conv                   BPU  id(0)     HzSQuantizedConv           0.992670           1.15481    int8      
/model.16/m.0/m/m.0/cv1/act/Mul                     BPU  id(0)     HzLut                      0.987919           5.2075     int8      
/model.16/m.0/m/m.0/cv2/conv/Conv                   BPU  id(0)     HzSQuantizedConv           0.990404           4.00224    int8      
/model.16/m.0/m/m.0/cv2/act/Mul                     BPU  id(0)     HzLut                      0.985384           6.39215    int8      
/model.16/m.0/m/m.0/Add                             BPU  id(0)     HzSElementwiseAdd          0.985627           1.15481    int8      
/model.16/m.0/m/m.1/cv1/conv/Conv                   BPU  id(0)     HzSQuantizedConv           0.990628           4.8667     int8      
/model.16/m.0/m/m.1/cv1/act/Mul                     BPU  id(0)     HzLut                      0.988135           6.04507    int8      
/model.16/m.0/m/m.1/cv2/conv/Conv                   BPU  id(0)     HzSQuantizedConv           0.986813           4.24945    int8      
/model.16/m.0/m/m.1/cv2/act/Mul                     BPU  id(0)     HzLut                      0.987800           10.5763    int8      
/model.16/m.0/m/m.1/Add                             BPU  id(0)     HzSElementwiseAdd          0.987854           4.8667     int8      
/model.16/m.0/Concat                                BPU  id(0)     Concat                     0.986582           10.2756    int8      
/model.16/m.0/cv3/conv/Conv                         BPU  id(0)     HzSQuantizedConv           0.987478           10.2756    int8      
/model.16/m.0/cv3/act/Mul                           BPU  id(0)     HzLut                      0.976658           7.10816    int8      
/model.16/Split_output_0_calibrated_Requantize      BPU  id(0)     HzRequantize               --                 --         int8      
/model.16/Split_output_1_calibrated_Requantize      BPU  id(0)     HzRequantize               --                 --         int8      
/model.16/Concat                                    BPU  id(0)     Concat                     0.989627           2.70114    int8      
/model.16/cv2/conv/Conv                             BPU  id(0)     HzSQuantizedConv           0.991644           3.47766    int8      
/model.16/cv2/act/Mul                               BPU  id(0)     HzLut                      0.990121           8.07615    int8      
/model.17/conv/Conv                                 BPU  id(0)     HzSQuantizedConv           0.989022           3.45944    int8      
/model.23/cv2.0/cv2.0.0/conv/Conv                   BPU  id(0)     HzSQuantizedConv           0.988720           3.45944    int8      
/model.23/cv3.0/cv3.0.0/cv3.0.0.0/conv/Conv         BPU  id(0)     HzSQuantizedConv           0.992291           3.45944    int8      
/model.17/act/Mul                                   BPU  id(0)     HzLut                      0.977386           6.59337    int8      
/model.23/cv2.0/cv2.0.0/act/Mul                     BPU  id(0)     HzLut                      0.985188           8.557      int8      
/model.23/cv3.0/cv3.0.0/cv3.0.0.0/act/Mul           BPU  id(0)     HzLut                      0.991365           5.82545    int8      
/model.18/Concat                                    BPU  id(0)     Concat                     0.982981           5.36634    int8      
/model.23/cv2.0/cv2.0.1/conv/Conv                   BPU  id(0)     HzSQuantizedConv           0.983774           8.17024    int8      
/model.23/cv3.0/cv3.0.0/cv3.0.0.1/conv/Conv         BPU  id(0)     HzSQuantizedConv           0.980573           5.78093    int8      
/model.19/cv1/conv/Conv                             BPU  id(0)     HzSQuantizedConv           0.989142           5.36634    int8      
/model.23/cv2.0/cv2.0.1/act/Mul                     BPU  id(0)     HzLut                      0.985555           25.5971    int8      
/model.23/cv3.0/cv3.0.0/cv3.0.0.1/act/Mul           BPU  id(0)     HzLut                      0.978560           7.71945    int8      
/model.19/cv1/act/Mul                               BPU  id(0)     HzLut                      0.978135           7.71698    int8      
/model.23/cv2.0/cv2.0.2/Conv                        BPU  id(0)     HzSQuantizedConv           0.994190           22.388     int8      
/model.23/cv3.0/cv3.0.1/cv3.0.1.0/conv/Conv         BPU  id(0)     HzSQuantizedConv           0.980891           6.82588    int8      
/model.19/Split                                     BPU  id(0)     Split                      0.978910           7.32857    int8      
/model.19/m.0/cv1/conv/Conv                         BPU  id(0)     HzSQuantizedConv           0.990718           7.32857    int8      
/model.19/m.0/cv2/conv/Conv                         BPU  id(0)     HzSQuantizedConv           0.980376           7.32857    int8      
/model.23/cv3.0/cv3.0.1/cv3.0.1.0/act/Mul           BPU  id(0)     HzLut                      0.984099           9.44061    int8      
/model.23/cv3.0/cv3.0.1/cv3.0.1.1/conv/Conv         BPU  id(0)     HzSQuantizedConv           0.986676           7.19369    int8      
/model.19/m.0/cv1/act/Mul                           BPU  id(0)     HzLut                      0.989914           4.27451    int8      
/model.19/m.0/cv2/act/Mul                           BPU  id(0)     HzLut                      0.970810           7.98388    int8      
/model.19/m.0/m/m.0/cv1/conv/Conv                   BPU  id(0)     HzSQuantizedConv           0.990906           3.18927    int8      
/model.23/cv3.0/cv3.0.1/cv3.0.1.1/act/Mul           BPU  id(0)     HzLut                      0.987971           19.0528    int8      
/model.23/cv3.0/cv3.0.2/Conv                        BPU  id(0)     HzSQuantizedConv           0.999610           18.2995    int8      
/model.19/m.0/m/m.0/cv1/act/Mul                     BPU  id(0)     HzLut                      0.975895           6.91989    int8      
/model.23/Concat                                    BPU  id(0)     Concat                     0.998145           24.2492    int16     
/model.19/m.0/m/m.0/cv2/conv/Conv                   BPU  id(0)     HzSQuantizedConv           0.985772           5.0844     int8      
/model.23/Reshape                                   BPU  id(0)     Reshape                    0.998145           24.2492    int16     
/model.19/m.0/m/m.0/cv2/act/Mul                     BPU  id(0)     HzLut                      0.974666           7.65872    int8      
/model.19/m.0/m/m.0/Add                             BPU  id(0)     HzSElementwiseAdd          0.977309           3.18927    int8      
/model.19/m.0/m/m.1/cv1/conv/Conv                   BPU  id(0)     HzSQuantizedConv           0.989714           7.27595    int8      
/model.19/m.0/m/m.1/cv1/act/Mul                     BPU  id(0)     HzLut                      0.970970           7.61422    int8      
/model.19/m.0/m/m.1/cv2/conv/Conv                   BPU  id(0)     HzSQuantizedConv           0.979136           6.64711    int8      
/model.19/m.0/m/m.1/cv2/act/Mul                     BPU  id(0)     HzLut                      0.972467           12.8757    int8      
/model.19/m.0/m/m.1/Add                             BPU  id(0)     HzSElementwiseAdd          0.972535           7.27595    int8      
/model.19/m.0/Concat                                BPU  id(0)     Concat                     0.972421           12.7693    int8      
/model.19/m.0/cv3/conv/Conv                         BPU  id(0)     HzSQuantizedConv           0.976778           12.7693    int8      
/model.19/m.0/cv3/act/Mul                           BPU  id(0)     HzLut                      0.965923           12.0066    int8      
/model.19/Split_output_0_calibrated_Requantize      BPU  id(0)     HzRequantize               --                 --         int8      
/model.19/Split_output_1_calibrated_Requantize      BPU  id(0)     HzRequantize               --                 --         int8      
/model.19/Concat                                    BPU  id(0)     Concat                     0.970942           7.32857    int8      
/model.19/cv2/conv/Conv                             BPU  id(0)     HzSQuantizedConv           0.980509           7.36259    int8      
/model.19/cv2/act/Mul                               BPU  id(0)     HzLut                      0.968740           9.60828    int8      
/model.20/conv/Conv                                 BPU  id(0)     HzSQuantizedConv           0.981389           9.52471    int8      
/model.23/cv2.1/cv2.1.0/conv/Conv                   BPU  id(0)     HzSQuantizedConv           0.982915           9.52471    int8      
/model.23/cv3.1/cv3.1.0/cv3.1.0.0/conv/Conv         BPU  id(0)     HzSQuantizedConv           0.971279           9.52471    int8      
/model.20/act/Mul                                   BPU  id(0)     HzLut                      0.974258           6.8139     int8      
/model.23/cv2.1/cv2.1.0/act/Mul                     BPU  id(0)     HzLut                      0.975998           9.55699    int8      
/model.23/cv3.1/cv3.1.0/cv3.1.0.0/act/Mul           BPU  id(0)     HzLut                      0.969601           8.46081    int8      
/model.21/Concat                                    BPU  id(0)     Concat                     0.979098           5.68315    int8      
/model.23/cv2.1/cv2.1.1/conv/Conv                   BPU  id(0)     HzSQuantizedConv           0.981122           9.37754    int8      
/model.23/cv3.1/cv3.1.0/cv3.1.0.1/conv/Conv         BPU  id(0)     HzSQuantizedConv           0.977093           7.15585    int8      
/model.22/cv1/conv/Conv                             BPU  id(0)     HzSQuantizedConv           0.985039           5.68315    int8      
/model.23/cv2.1/cv2.1.1/act/Mul                     BPU  id(0)     HzLut                      0.982029           20.4932    int8      
/model.23/cv3.1/cv3.1.0/cv3.1.0.1/act/Mul           BPU  id(0)     HzLut                      0.971847           8.48387    int8      
/model.22/cv1/act/Mul                               BPU  id(0)     HzLut                      0.981744           5.99525    int8      
/model.23/cv2.1/cv2.1.2/Conv                        BPU  id(0)     HzSQuantizedConv           0.992312           20.2963    int8      
/model.23/cv3.1/cv3.1.1/cv3.1.1.0/conv/Conv         BPU  id(0)     HzSQuantizedConv           0.974499           7.94545    int8      
/model.22/Split                                     BPU  id(0)     Split                      0.982677           5.77492    int8      
/model.22/m.0/cv1/conv/Conv                         BPU  id(0)     HzSQuantizedConv           0.989638           5.77492    int8      
/model.22/m.0/cv2/conv/Conv                         BPU  id(0)     HzSQuantizedConv           0.984474           5.77492    int8      
/model.23/cv3.1/cv3.1.1/cv3.1.1.0/act/Mul           BPU  id(0)     HzLut                      0.976625           11.0047    int8      
/model.23/cv3.1/cv3.1.1/cv3.1.1.1/conv/Conv         BPU  id(0)     HzSQuantizedConv           0.980663           9.9034     int8      
/model.22/m.0/cv1/act/Mul                           BPU  id(0)     HzLut                      0.986369           3.81192    int8      
/model.22/m.0/cv2/act/Mul                           BPU  id(0)     HzLut                      0.981457           5.10639    int8      
/model.22/m.0/m/m.0/cv1/conv/Conv                   BPU  id(0)     HzSQuantizedConv           0.992100           2.14707    int8      
/model.23/cv3.1/cv3.1.1/cv3.1.1.1/act/Mul           BPU  id(0)     HzLut                      0.983421           16.8918    int8      
/model.23/cv3.1/cv3.1.2/Conv                        BPU  id(0)     HzSQuantizedConv           0.999540           16.608     int8      
/model.22/m.0/m/m.0/cv1/act/Mul                     BPU  id(0)     HzLut                      0.987676           4.82011    int8      
/model.23/Concat_1                                  BPU  id(0)     Concat                     0.997271           24.2492    int16     
/model.22/m.0/m/m.0/cv2/conv/Conv                   BPU  id(0)     HzSQuantizedConv           0.995446           3.83068    int8      
/model.23/Reshape_1                                 BPU  id(0)     Reshape                    0.997271           24.2492    int16     
/model.22/m.0/m/m.0/cv2/act/Mul                     BPU  id(0)     HzLut                      0.993676           5.29034    int8      
/model.22/m.0/m/m.0/Add                             BPU  id(0)     HzSElementwiseAdd          0.992092           2.14707    int8      
/model.22/m.0/m/m.1/cv1/conv/Conv                   BPU  id(0)     HzSQuantizedConv           0.996907           4.73458    int8      
/model.22/m.0/m/m.1/cv1/act/Mul                     BPU  id(0)     HzLut                      0.993510           4.64762    int8      
/model.22/m.0/m/m.1/cv2/conv/Conv                   BPU  id(0)     HzSQuantizedConv           0.996690           3.93651    int8      
/model.22/m.0/m/m.1/cv2/act/Mul                     BPU  id(0)     HzLut                      0.996251           6.46214    int8      
/model.22/m.0/m/m.1/Add                             BPU  id(0)     HzSElementwiseAdd          0.995055           4.73458    int8      
/model.22/m.0/Concat                                BPU  id(0)     Concat                     0.993324           6.39085    int8      
/model.22/m.0/cv3/conv/Conv                         BPU  id(0)     HzSQuantizedConv           0.995405           6.39085    int8      
/model.22/m.0/cv3/act/Mul                           BPU  id(0)     HzLut                      0.994633           5.50092    int8      
/model.22/Split_output_0_calibrated_Requantize      BPU  id(0)     HzRequantize               --                 --         int8      
/model.22/Split_output_1_calibrated_Requantize      BPU  id(0)     HzRequantize               --                 --         int8      
/model.22/Concat                                    BPU  id(0)     Concat                     0.988932           5.77492    int8      
/model.22/cv2/conv/Conv                             BPU  id(0)     HzSQuantizedConv           0.994583           5.64372    int8      
/model.22/cv2/act/Mul                               BPU  id(0)     HzLut                      0.992129           5.8584     int8      
/model.23/cv2.2/cv2.2.0/conv/Conv                   BPU  id(0)     HzSQuantizedConv           0.997616           5.75717    int8      
/model.23/cv3.2/cv3.2.0/cv3.2.0.0/conv/Conv         BPU  id(0)     HzSQuantizedConv           0.995095           5.75717    int8      
/model.23/cv2.2/cv2.2.0/act/Mul                     BPU  id(0)     HzLut                      0.995666           4.9207     int8      
/model.23/cv3.2/cv3.2.0/cv3.2.0.0/act/Mul           BPU  id(0)     HzLut                      0.993215           4.69073    int8      
/model.23/cv2.2/cv2.2.1/conv/Conv                   BPU  id(0)     HzSQuantizedConv           0.996364           4.42371    int8      
/model.23/cv3.2/cv3.2.0/cv3.2.0.1/conv/Conv         BPU  id(0)     HzSQuantizedConv           0.996598           4.49101    int8      
/model.23/cv2.2/cv2.2.1/act/Mul                     BPU  id(0)     HzLut                      0.995875           14.9887    int8      
/model.23/cv3.2/cv3.2.0/cv3.2.0.1/act/Mul           BPU  id(0)     HzLut                      0.996465           5.68772    int8      
/model.23/cv2.2/cv2.2.2/Conv                        BPU  id(0)     HzSQuantizedConv           0.998494           14.9887    int8      
/model.23/cv3.2/cv3.2.1/cv3.2.1.0/conv/Conv         BPU  id(0)     HzSQuantizedConv           0.997867           4.50133    int8      
/model.23/cv3.2/cv3.2.1/cv3.2.1.0/act/Mul           BPU  id(0)     HzLut                      0.997746           5.99539    int8      
/model.23/cv3.2/cv3.2.1/cv3.2.1.1/conv/Conv         BPU  id(0)     HzSQuantizedConv           0.998814           5.9805     int8      
/model.23/cv3.2/cv3.2.1/cv3.2.1.1/act/Mul           BPU  id(0)     HzLut                      0.999163           7.51983    int8      
/model.23/cv3.2/cv3.2.2/Conv                        BPU  id(0)     HzSQuantizedConv           0.999992           7.51575    int8      
/model.23/Concat_2                                  BPU  id(0)     Concat                     0.999773           24.2492    int16     
/model.23/Reshape_2                                 BPU  id(0)     Reshape                    0.999773           24.2492    int16     
/model.23/Concat_3                                  BPU  id(0)     Concat                     0.998017           24.2492    int16     
/model.23/Split                                     BPU  id(0)     Split                      0.993701           24.2492    int16     
/model.23/dfl/Reshape                               BPU  id(0)     Reshape                    0.993701           24.2492    int16     
/model.23/Sigmoid                                   BPU  id(0)     HzLut2Layer                0.991018           24.2492    int16     
...23/dfl/Softmax_reducemax_FROM_QUANTIZED_SOFTMAX  BPU  id(0)     HzQuantizedReduceMax       0.996419           24.2492    int16     
/model.23/dfl/Softmax_sub_FROM_QUANTIZED_SOFTMAX    BPU  id(0)     HzSElementwiseSub          0.947156           24.2492    int16     
/model.23/dfl/Softmax_exp_FROM_QUANTIZED_SOFTMAX    BPU  id(0)     HzLut                      0.979501           5.64418    int8      
...23/dfl/Softmax_reducesum_FROM_QUANTIZED_SOFTMAX  CPU  --        ReduceSum                  0.987553           1.0        float     
...3/dfl/Softmax_reciprocal_FROM_QUANTIZED_SOFTMAX  BPU  id(1)     HzLut                      0.992624           11.3959    int8      
/model.23/dfl/Softmax_mul_FROM_QUANTIZED_SOFTMAX    BPU  id(1)     HzSElementwiseMul          0.983926           1.0        int8      
/model.23/dfl/conv/Conv                             BPU  id(1)     HzSQuantizedConv           0.995213           0.92482    int8      
/model.23/Slice                                     BPU  id(1)     Slice                      0.995184           10.0742    int8      
/model.23/Slice_1                                   BPU  id(1)     Slice                      0.995248           10.0742    int8      
/model.23/Sub                                       BPU  id(1)     HzSElementwiseSub          0.999938           79.5       int8      
/model.23/Add_1                                     BPU  id(1)     HzSElementwiseAdd          0.999952           79.5       int8      
/model.23/Add_2                                     BPU  id(1)     HzSElementwiseAdd          0.999965           79.3058    int8      
/model.23/Sub_1                                     BPU  id(1)     HzSElementwiseSub          0.995631           84.1767    int8      
/model.23/Div_1_mul                                 BPU  id(1)     HzSQuantizedConv           0.999965           159.983    int8      
/model.23/Concat_4                                  BPU  id(1)     Concat                     0.999844           79.9084    int8      
/model.23/Mul_2                                     BPU  id(1)     HzSElementwiseMul          0.999729           79.9084    int8      
/model.23/Sigmoid_output_0_calibrated_Requantize    BPU  id(0)     HzRequantize               --                 --         int16     
/model.23/Concat_5                                  BPU  id(1)     Concat                     0.999729           646.355    int16
2025-08-30 21:39:47,916 INFO The quantized model output:
=============================================================================
Output      Cosine Similarity  L1 Distance  L2 Distance  Chebyshev Distance  
-----------------------------------------------------------------------------
output0     0.999729           1.905012     0.015577     112.867523
2025-08-30 21:39:47,922 INFO End to Horizon NN Model Convert.
2025-08-30 21:39:47,946 INFO start convert to *.bin file....
2025-08-30 21:39:47,981 INFO ONNX model output num : 1
2025-08-30 21:39:47,989 INFO ############# model deps info #############
2025-08-30 21:39:47,989 INFO hb_mapper version   : 1.24.3
2025-08-30 21:39:47,989 INFO hbdk version        : 3.49.15
2025-08-30 21:39:47,989 INFO hbdk runtime version: 3.15.55.0
2025-08-30 21:39:47,989 INFO horizon_nn version  : 1.1.0
2025-08-30 21:39:47,989 INFO ############# model_parameters info #############
2025-08-30 21:39:47,989 INFO onnx_model          : /open_explorer/yolo11m/best.onnx
2025-08-30 21:39:47,989 INFO BPU march           : bayes-e
2025-08-30 21:39:47,989 INFO layer_out_dump      : False
2025-08-30 21:39:47,990 INFO log_level           : DEBUG
2025-08-30 21:39:47,990 INFO working dir         : /open_explorer/yolo11m/.temporary_workspace/bpu_model_output
2025-08-30 21:39:47,990 INFO output_model_file_prefix: best_bayese_640x640_nv12
2025-08-30 21:39:47,990 INFO ############# input_parameters info #############
2025-08-30 21:39:47,990 INFO ------------------------------------------
2025-08-30 21:39:47,990 INFO ---------input info : images ---------
2025-08-30 21:39:47,990 INFO input_name          : images
2025-08-30 21:39:47,990 INFO input_type_rt       : nv12
2025-08-30 21:39:47,990 INFO input_space&range   : regular
2025-08-30 21:39:47,990 INFO input_layout_rt     : None
2025-08-30 21:39:47,990 INFO input_type_train    : rgb
2025-08-30 21:39:47,990 INFO input_layout_train  : NCHW
2025-08-30 21:39:47,990 INFO norm_type           : data_scale
2025-08-30 21:39:47,990 INFO input_shape         : 1x3x640x640
2025-08-30 21:39:47,990 INFO scale_value         : 0.003921568627451,
2025-08-30 21:39:47,990 INFO cal_data_dir        : /open_explorer/yolo11m/.temporary_workspace/.calibration_data_temporary_folder
2025-08-30 21:39:47,990 INFO cal_data_type       : float32
2025-08-30 21:39:47,990 INFO ---------input info : images end -------
2025-08-30 21:39:47,991 INFO ------------------------------------------
2025-08-30 21:39:47,991 INFO ############# calibration_parameters info #############
2025-08-30 21:39:47,991 INFO preprocess_on       : False
2025-08-30 21:39:47,991 INFO calibration_type:   : default
2025-08-30 21:39:47,991 INFO optimization        : set_Softmax_input_int8,set_Softmax_output_int8;
2025-08-30 21:39:47,991 INFO per_channel         : False
2025-08-30 21:39:47,991 INFO ############# compiler_parameters info #############
2025-08-30 21:39:47,991 INFO debug               : True
2025-08-30 21:39:47,991 INFO compile_mode        : latency
2025-08-30 21:39:47,991 INFO hbdk_pass_through_params: --O3 --debug --core-num 1 --fast --jobs 16
2025-08-30 21:39:47,991 INFO input-source        : {'images': 'pyramid', '_default_value': 'ddr'}
2025-08-30 21:39:48,016 INFO Convert to runtime bin file successfully!
2025-08-30 21:39:48,016 INFO End Model Convert
[MZOO] [21:39:48.714] [INFO] Model conversion completed successfully!
[MZOO] [21:39:48.714] [INFO] Looking for output file: /open_explorer/yolo11m/.temporary_workspace/bpu_model_output/best_bayese_640x640_nv12.bin
[MZOO] [21:39:48.714] [INFO] Output file moved to: /open_explorer/yolo11m/best_bayese_640x640_nv12.bin
[MZOO] [21:39:48.715] [INFO] hb_mapper log moved to: /open_explorer/yolo11m/hb_mapper_makertbin.log
[MZOO] [21:39:48.715] [INFO] Restored working directory to: /open_explorer/yolo11m
[MZOO] [21:39:48.715] [INFO] Cleaning up...
[MZOO] [21:39:48.800] [INFO] Temporary files cleaned up.
[MZOO] [21:39:48.800] [INFO] Conversion completed successfully!
[MZOO] [21:39:48.800] [INFO] Output file: /open_explorer/yolo11m/best_bayese_640x640_nv12.bin
[MZOO] [21:39:48.800] [INFO] Mapper log: /open_explorer/yolo11m/hb_mapper_makertbin.log
2

Hi swiftfile, welcome to our forum community!

It turns out that after training the YOLO model, at the time you convert the model into ONNX, you have to use export_monkey_patch.py to revise the detector head to make it BPU adaptable.
Please take a look at this link: rdk_model_zoo/demos/Vision/ultralytics_YOLO at main · D-Robotics/rdk_model_zoo · GitHub

image
image1098×1052 117 KB

Let me know after you try that.

D-robotics

1 Like
3

Hi d-robot1 :smiley:
Thank you for your reply!
I followed your procedure as below and I converted onnx to bin file with only BPU process.

  1. I ran python3 export_monkey_patch.py --pt best.pt
  2. python3 rdk_model_zoo/demos/Vision/ultralytics_YOLO/x86/mapper.py --onnx best.onnx --cal-images Robomaster.v3i.yolov11/train/images/ this script modified to my converted weight,best.onnx.
  3. The following image is the output of hb_mapper checker --model-type onnx --model ./best.onnx

best_bayese_640x640_nv12
best_bayese_640x640_nv123603×531 67.9 KB

The graph is similar to downloaded weight of model zoo.
However, I’m not sure whether this is valid for image inference.
Could you share your opinion?
Also, when I tried running the inference program(ros2 launch dnn_node_example dnn_node_example.launch.py). it did not execute properly. :sad_but_relieved_face:

This is the log from when I convert the weight file.

root@35ede7ae0dad:/open_explorer# python3 rdk_model_zoo/demos/Vision/ultralytics_YOLO/x86/mapper.py --onnx best.onnx --cal-images Robomaster.v3i.yolov11/train/images/
[MZOO] [19:12:19.323] [INFO] Namespace(cal_images='Robomaster.v3i.yolov11/train/images/', onnx='best.onnx', output_dir='.', quantized='int8', jobs=16, optimize_level='O3', cal_sample=True, cal_sample_num=20, save_cache=False, cal='.calibration_data_temporary_folder', ws='.temporary_workspace')
[MZOO] [19:12:19.324] [INFO] Output directory set to ONNX file directory: /open_explorer
[MZOO] [19:12:19.324] [INFO] Resolved paths:
[MZOO] [19:12:19.324] [INFO]   ONNX model: /open_explorer/best.onnx
[MZOO] [19:12:19.324] [INFO]   Calibration images: /open_explorer/Robomaster.v3i.yolov11/train/images
[MZOO] [19:12:19.324] [INFO]   Output directory: /open_explorer
[MZOO] [19:12:19.324] [INFO]   Workspace: /open_explorer/.temporary_workspace
[MZOO] [19:12:21.344] [INFO] hb_mapper is available and working.
[MZOO] [19:12:21.345] [INFO] Loading ONNX model from: /open_explorer/best.onnx
[MZOO] [19:12:21.514] [DEBUG] Model has a single input.
[MZOO] [19:12:21.515] [DEBUG] Input data type is float32 (tensor(float)).
[MZOO] [19:12:21.515] [DEBUG] NCHW check success.
[MZOO] [19:12:21.515] [DEBUG] ONNX Runtime session released.
[MZOO] [19:12:21.515] [INFO] Sampling enabled. Sampled 20 images from 50 total images.
[MZOO] [19:12:21.515] [INFO] Workspace '/open_explorer/.temporary_workspace' created successfully.
[MZOO] [19:12:21.515] [INFO] Configuration file created: /open_explorer/.temporary_workspace/config.yaml
[MZOO] [19:12:21.515] [INFO] Preparing calibration data...
[MZOO] [19:12:21.996] [INFO] Calibration data has been successfully generated.
[MZOO] [19:12:21.996] [INFO] Changed working directory to: /open_explorer/.temporary_workspace
[MZOO] [19:12:21.996] [INFO] Executing model conversion...
[MZOO] [19:12:21.996] [INFO] Command: hb_mapper makertbin --config config.yaml --model-type onnx
2025-09-14 19:12:23,677 INFO log will be stored in /open_explorer/.temporary_workspace/hb_mapper_makertbin.log
2025-09-14 19:12:23,677 INFO Start hb_mapper....
2025-09-14 19:12:23,678 INFO hbdk version 3.49.15
2025-09-14 19:12:23,678 INFO horizon_nn version 1.1.0
2025-09-14 19:12:23,678 INFO hb_mapper version 1.24.3
2025-09-14 19:12:23,678 INFO Start Model Convert....
2025-09-14 19:12:23,682 INFO Using onnx model file: /open_explorer/best.onnx
2025-09-14 19:12:23,728 INFO Model has 1 inputs according to model file
2025-09-14 19:12:23,728 INFO working_dir does not exist. Creating working_dir: /open_explorer/.temporary_workspace/bpu_model_output
2025-09-14 19:12:23,729 INFO Model name not given in yaml_file, using model name from model file: ['images']
2025-09-14 19:12:23,729 INFO Model input shape not given in yaml_file, using shape from model file: [[1, 3, 640, 640]]
2025-09-14 19:12:23,729 INFO nv12 input type rt received.
2025-09-14 19:12:23,729 WARNING The calibration dir name suffix is not the same as the value float32 of the parameter cal_data_type, the parameter setting will prevail
2025-09-14 19:12:23,729 INFO custom_op does not exist, skipped
2025-09-14 19:12:23,729 WARNING Input node images's input_source not set, it will be set to pyramid by default
2025-09-14 19:12:23,732 INFO *******************************************
2025-09-14 19:12:23,732 INFO First calibration picture name: -AresVs-_BO2_2_253_jpg.rf.6e5a239db406eceee1e4866c6fe4bcc5.jpg.rgbchw
2025-09-14 19:12:23,732 INFO First calibration picture md5:
c33235f5ea432dce8d2a66dca4caf841  /open_explorer/.temporary_workspace/.calibration_data_temporary_folder/-AresVs-_BO2_2_253_jpg.rf.6e5a239db406eceee1e4866c6fe4bcc5.jpg.rgbchw
2025-09-14 19:12:23,752 INFO *******************************************
2025-09-14 19:12:25,340 INFO Start to Horizon NN Model Convert.
2025-09-14 19:12:25,393 INFO Loading horizon_nn debug methods:set()
2025-09-14 19:12:25,394 INFO The activation calibration parameters:
    calibration_type:     ['max', 'kl']
    max_percentile:       [0.99995, 1.0]
    per_channel:          [True, False]
    asymmetric:           [True, False]
The modelwise search parameters:
    similarity:           0.995
    metric:               cosine-similarity
The input of all Softmax nodes are set to : int8
2025-09-14 19:12:25,394 INFO input images is from pyramid. Its layout is set to NHWC
2025-09-14 19:12:25,394 INFO The specified model compilation architecture: bayes-e.
2025-09-14 19:12:25,394 INFO The specified model compilation optimization parameters: [].
2025-09-14 19:12:25,394 INFO Start to prepare the onnx model.
2025-09-14 19:12:25,463 INFO Input ONNX Model Information:
ONNX IR version:          6
Opset version:            ['ai.onnx v11', 'horizon v1']
Producer:                 pytorch v1.13.0
Domain:                   None
Version:                  None
Graph input:
    images:               shape=[1, 3, 640, 640], dtype=FLOAT32
Graph output:
    output0:              shape=[1, 80, 80, 5], dtype=FLOAT32
    480:                  shape=[1, 80, 80, 64], dtype=FLOAT32
    494:                  shape=[1, 40, 40, 5], dtype=FLOAT32
    502:                  shape=[1, 40, 40, 64], dtype=FLOAT32
    516:                  shape=[1, 20, 20, 5], dtype=FLOAT32
    524:                  shape=[1, 20, 20, 64], dtype=FLOAT32
2025-09-14 19:12:26,024 INFO End to prepare the onnx model.
2025-09-14 19:12:26,096 INFO Saving model to: best_bayese_640x640_nv12_original_float_model.onnx.
2025-09-14 19:12:26,096 INFO Start to optimize the onnx model.
2025-09-14 19:12:26,858 INFO End to optimize the onnx model.
2025-09-14 19:12:26,923 INFO Saving model to: best_bayese_640x640_nv12_optimized_float_model.onnx.
2025-09-14 19:12:26,924 INFO Start to calibrate the model.
2025-09-14 19:12:27,187 INFO There are 20 samples in the data set.
2025-09-14 19:12:27,187 INFO Run calibration model with modelwise search method.
2025-09-14 19:12:27,755 INFO Calibration using batch 8
calibration in progress:   0%|                                          | 0/3 [00:00<?, ?it/s]2025-09-14 19:12:34.484496877 [E:onnxruntime:, sequential_executor.cc:183 Execute] Non-zero status code returned while running Reshape node. Name:'/model.10/m/m.0/attn/Reshape' Status Message: /home/jenkins/agent/workspace/model_convert_hmct1_ut_cpu_1.1.0/onnxruntime/onnxruntime/core/providers/cpu/tensor/reshape_helper.h:43 onnxruntime::ReshapeHelper::ReshapeHelper(const onnxruntime::TensorShape&, std::vector<long int>&) gsl::narrow_cast<int64_t>(input_shape.Size()) == size was false. The input tensor cannot be reshaped to the requested shape. Input shape:{8,512,20,20}, requested shape:{1,4,128,400}

calibration in progress:   0%|                                          | 0/3 [00:06<?, ?it/s]
2025-09-14 19:12:34,485 INFO Reset batch_size=1 and execute forward again...
calibration in progress: 100%|████████████████████████████████| 20/20 [01:12<00:00,  3.62s/it]
2025-09-14 19:14:04,412 INFO Select max-percentile:percentile=0.99995 method.
2025-09-14 19:14:05,587 INFO End to calibrate the model.
2025-09-14 19:14:05,749 INFO Saving model to: best_bayese_640x640_nv12_calibrated_model.onnx.
2025-09-14 19:14:05,749 INFO Start to quantize the model.
2025-09-14 19:14:08,466 INFO End to quantize the model.
2025-09-14 19:14:08,529 INFO Saving model to: best_bayese_640x640_nv12_quantized_model.onnx.
2025-09-14 19:14:08,529 INFO Start to compile the model with march bayes-e.
2025-09-14 19:14:09,059 INFO Compile submodel: torch_jit_subgraph_0
2025-09-14 19:14:09,104 INFO hbdk-cc parameters:['--O3', '--debug', '--core-num', '1', '--fast', '--jobs', '16', '--input-layout', 'NHWC', '--output-layout', 'NHWC', '--input-source', 'pyramid']
[==================================================] 100%
2025-09-14 19:19:04,507 INFO consumed time 295.374
2025-09-14 19:19:04,705 INFO FPS=79.44, latency = 12587.7 us, DDR = 40253952 bytes   (see torch_jit_subgraph_0.html)
2025-09-14 19:19:04,916 INFO End to compile the model with march bayes-e.
2025-09-14 19:19:09,925 INFO The main quantized node information:
======================================================================================================================================
Node                                                ON   Subgraph  Type                       Cosine Similarity  Threshold  DataType  
--------------------------------------------------------------------------------------------------------------------------------------
HZ_PREPROCESS_FOR_images                            BPU  id(0)     HzSQuantizedPreprocess     0.999968           127.0      int8      
/model.0/conv/Conv                                  BPU  id(0)     HzSQuantizedConv           0.998878           1.03771    int8      
/model.0/act/Mul                                    BPU  id(0)     HzLut                      0.997506           50.6915    int8      
/model.1/conv/Conv                                  BPU  id(0)     HzSQuantizedConv           0.996750           45.5221    int8      
/model.1/act/Mul                                    BPU  id(0)     HzLut                      0.995413           54.6037    int8      
/model.2/cv1/conv/Conv                              BPU  id(0)     HzSQuantizedConv           0.995624           43.0455    int8      
/model.2/cv1/act/Mul                                BPU  id(0)     HzLut                      0.993788           38.5107    int8      
/model.2/Split                                      BPU  id(0)     Split                      0.993585           10.719     int8      
/model.2/m.0/cv1/conv/Conv                          BPU  id(0)     HzSQuantizedConv           0.996310           10.719     int8      
/model.2/m.0/cv1/act/Mul                            BPU  id(0)     HzLut                      0.996290           7.87008    int8      
/model.2/m.0/cv2/conv/Conv                          BPU  id(0)     HzSQuantizedConv           0.991041           6.23633    int8      
/model.2/m.0/cv2/act/Mul                            BPU  id(0)     HzLut                      0.993388           11.0673    int8      
/model.2/m.0/Add                                    BPU  id(0)     HzSElementwiseAdd          0.995010           10.719     int8      
/model.2/Split_output_0_calibrated_Requantize       BPU  id(0)     HzRequantize               --                 --         int8      
/model.2/Split_output_1_calibrated_Requantize       BPU  id(0)     HzRequantize               --                 --         int8      
/model.2/Concat                                     BPU  id(0)     Concat                     0.994629           10.719     int8      
/model.2/cv2/conv/Conv                              BPU  id(0)     HzSQuantizedConv           0.985429           11.7718    int8      
/model.2/cv2/act/Mul                                BPU  id(0)     HzLut                      0.978021           11.6896    int8      
/model.3/conv/Conv                                  BPU  id(0)     HzSQuantizedConv           0.992106           6.35667    int8      
/model.3/act/Mul                                    BPU  id(0)     HzLut                      0.994263           7.55898    int8      
/model.4/cv1/conv/Conv                              BPU  id(0)     HzSQuantizedConv           0.993048           4.12643    int8      
/model.4/cv1/act/Mul                                BPU  id(0)     HzLut                      0.992933           7.57814    int8      
/model.4/Split                                      BPU  id(0)     Split                      0.993160           4.02228    int8      
/model.4/m.0/cv1/conv/Conv                          BPU  id(0)     HzSQuantizedConv           0.992788           4.02228    int8      
/model.4/m.0/cv1/act/Mul                            BPU  id(0)     HzLut                      0.994264           3.71548    int8      
/model.4/m.0/cv2/conv/Conv                          BPU  id(0)     HzSQuantizedConv           0.993583           2.93807    int8      
/model.4/m.0/cv2/act/Mul                            BPU  id(0)     HzLut                      0.994557           5.60615    int8      
/model.4/m.0/Add                                    BPU  id(0)     HzSElementwiseAdd          0.995405           4.02228    int8      
/model.4/Split_output_0_calibrated_Requantize       BPU  id(0)     HzRequantize               --                 --         int8      
/model.4/Split_output_1_calibrated_Requantize       BPU  id(0)     HzRequantize               --                 --         int8      
/model.4/Concat                                     BPU  id(0)     Concat                     0.994454           4.02228    int8      
/model.4/cv2/conv/Conv                              BPU  id(0)     HzSQuantizedConv           0.992886           5.09451    int8      
/model.4/cv2/act/Mul                                BPU  id(0)     HzLut                      0.985704           6.0629     int8      
/model.5/conv/Conv                                  BPU  id(0)     HzSQuantizedConv           0.993710           3.21265    int8      
/model.5/act/Mul                                    BPU  id(0)     HzLut                      0.989691           6.49808    int8      
/model.6/cv1/conv/Conv                              BPU  id(0)     HzSQuantizedConv           0.990470           3.52733    int8      
/model.6/cv1/act/Mul                                BPU  id(0)     HzLut                      0.985138           6.83378    int8      
/model.6/Split                                      BPU  id(0)     Split                      0.983869           4.41536    int8      
/model.6/m.0/cv1/conv/Conv                          BPU  id(0)     HzSQuantizedConv           0.994145           4.41536    int8      
/model.6/m.0/cv1/act/Mul                            BPU  id(0)     HzLut                      0.994008           3.95392    int8      
/model.6/m.0/m/m.0/cv1/conv/Conv                    BPU  id(0)     HzSQuantizedConv           0.995018           2.90757    int8      
/model.6/m.0/m/m.0/cv1/act/Mul                      BPU  id(0)     HzLut                      0.987649           5.70533    int8      
/model.6/m.0/m/m.0/cv2/conv/Conv                    BPU  id(0)     HzSQuantizedConv           0.992511           3.71571    int8      
/model.6/m.0/m/m.0/cv2/act/Mul                      BPU  id(0)     HzLut                      0.990338           4.79329    int8      
/model.6/m.0/m/m.0/Add                              BPU  id(0)     HzSElementwiseAdd          0.992825           2.90757    int8      
/model.6/m.0/m/m.1/cv1/conv/Conv                    BPU  id(0)     HzSQuantizedConv           0.995417           4.09336    int8      
/model.6/m.0/m/m.1/cv1/act/Mul                      BPU  id(0)     HzLut                      0.990393           5.02906    int8      
/model.6/m.0/m/m.1/cv2/conv/Conv                    BPU  id(0)     HzSQuantizedConv           0.994454           3.72792    int8      
/model.6/m.0/m/m.1/cv2/act/Mul                      BPU  id(0)     HzLut                      0.994649           7.56768    int8      
/model.6/m.0/m/m.1/Add                              BPU  id(0)     HzSElementwiseAdd          0.995423           4.09336    int8      
/model.6/m.0/cv2/conv/Conv                          BPU  id(0)     HzSQuantizedConv           0.987916           4.41536    int8      
/model.6/m.0/cv2/act/Mul                            BPU  id(0)     HzLut                      0.983996           6.88459    int8      
/model.6/m.0/Concat                                 BPU  id(0)     Concat                     0.994025           6.85606    int8      
/model.6/m.0/cv3/conv/Conv                          BPU  id(0)     HzSQuantizedConv           0.991721           6.85606    int8      
/model.6/m.0/cv3/act/Mul                            BPU  id(0)     HzLut                      0.986447           6.52048    int8      
/model.6/Split_output_0_calibrated_Requantize       BPU  id(0)     HzRequantize               --                 --         int8      
/model.6/Split_output_1_calibrated_Requantize       BPU  id(0)     HzRequantize               --                 --         int8      
/model.6/Concat                                     BPU  id(0)     Concat                     0.985527           4.41536    int8      
/model.6/cv2/conv/Conv                              BPU  id(0)     HzSQuantizedConv           0.993509           4.80497    int8      
/model.6/cv2/act/Mul                                BPU  id(0)     HzLut                      0.984172           6.41841    int8      
/model.7/conv/Conv                                  BPU  id(0)     HzSQuantizedConv           0.992719           4.34763    int8      
/model.7/act/Mul                                    BPU  id(0)     HzLut                      0.985353           6.03854    int8      
/model.8/cv1/conv/Conv                              BPU  id(0)     HzSQuantizedConv           0.991257           4.36116    int8      
/model.8/cv1/act/Mul                                BPU  id(0)     HzLut                      0.983894           7.27839    int8      
/model.8/Split                                      BPU  id(0)     Split                      0.983401           6.07786    int8      
/model.8/m.0/cv1/conv/Conv                          BPU  id(0)     HzSQuantizedConv           0.992371           6.07786    int8      
/model.8/m.0/cv1/act/Mul                            BPU  id(0)     HzLut                      0.992040           6.07621    int8      
/model.8/m.0/m/m.0/cv1/conv/Conv                    BPU  id(0)     HzSQuantizedConv           0.995299           4.97721    int8      
/model.8/m.0/m/m.0/cv1/act/Mul                      BPU  id(0)     HzLut                      0.988843           6.20863    int8      
/model.8/m.0/m/m.0/cv2/conv/Conv                    BPU  id(0)     HzSQuantizedConv           0.993677           5.09723    int8      
/model.8/m.0/m/m.0/cv2/act/Mul                      BPU  id(0)     HzLut                      0.991113           6.41362    int8      
/model.8/m.0/m/m.0/Add                              BPU  id(0)     HzSElementwiseAdd          0.991713           4.97721    int8      
/model.8/m.0/m/m.1/cv1/conv/Conv                    BPU  id(0)     HzSQuantizedConv           0.996195           6.51029    int8      
/model.8/m.0/m/m.1/cv1/act/Mul                      BPU  id(0)     HzLut                      0.992408           7.6371     int8      
/model.8/m.0/m/m.1/cv2/conv/Conv                    BPU  id(0)     HzSQuantizedConv           0.995172           4.58976    int8      
/model.8/m.0/m/m.1/cv2/act/Mul                      BPU  id(0)     HzLut                      0.995162           9.16626    int8      
/model.8/m.0/m/m.1/Add                              BPU  id(0)     HzSElementwiseAdd          0.994397           6.51029    int8      
/model.8/m.0/cv2/conv/Conv                          BPU  id(0)     HzSQuantizedConv           0.987166           6.07786    int8      
/model.8/m.0/cv2/act/Mul                            BPU  id(0)     HzLut                      0.983643           6.96721    int8      
/model.8/m.0/Concat                                 BPU  id(0)     Concat                     0.993067           9.59859    int8      
/model.8/m.0/cv3/conv/Conv                          BPU  id(0)     HzSQuantizedConv           0.994112           9.59859    int8      
/model.8/m.0/cv3/act/Mul                            BPU  id(0)     HzLut                      0.991129           6.32045    int8      
/model.8/Split_output_0_calibrated_Requantize       BPU  id(0)     HzRequantize               --                 --         int8      
/model.8/Split_output_1_calibrated_Requantize       BPU  id(0)     HzRequantize               --                 --         int8      
/model.8/Concat                                     BPU  id(0)     Concat                     0.986338           6.07786    int8      
/model.8/cv2/conv/Conv                              BPU  id(0)     HzSQuantizedConv           0.991122           5.85594    int8      
/model.8/cv2/act/Mul                                BPU  id(0)     HzLut                      0.987705           7.99929    int8      
/model.9/cv1/conv/Conv                              BPU  id(0)     HzSQuantizedConv           0.996130           6.72806    int8      
/model.9/cv1/act/Mul                                BPU  id(0)     HzLut                      0.995647           7.73082    int8      
/model.9/m/MaxPool                                  BPU  id(0)     HzQuantizedMaxPool         0.998026           11.9448    int8      
/model.9/m_1/MaxPool                                BPU  id(0)     HzQuantizedMaxPool         0.998188           11.9448    int8      
/model.9/m_2/MaxPool                                BPU  id(0)     HzQuantizedMaxPool         0.998159           11.9448    int8      
/model.9/Concat                                     BPU  id(0)     Concat                     0.997983           11.9448    int8      
/model.9/cv2/conv/Conv                              BPU  id(0)     HzSQuantizedConv           0.995244           11.9448    int8      
/model.9/cv2/act/Mul                                BPU  id(0)     HzLut                      0.986763           7.5331     int8      
/model.10/cv1/conv/Conv                             BPU  id(0)     HzSQuantizedConv           0.990575           3.87879    int8      
/model.10/cv1/act/Mul                               BPU  id(0)     HzLut                      0.986690           9.73169    int8      
/model.10/Split                                     BPU  id(0)     Split                      0.983220           6.50959    int8      
/model.10/m/m.0/attn/qkv/conv/Conv                  BPU  id(0)     HzSQuantizedConv           0.990654           6.50959    int8      
/model.10/m/m.0/attn/Reshape                        BPU  id(0)     Reshape                    0.990654           6.27205    int8      
/model.10/m/m.0/attn/Split                          BPU  id(0)     Split                      0.992509           6.27205    int8      
/model.10/m/m.0/attn/Transpose                      BPU  id(0)     Transpose                  0.992509           6.27205    int8      
/model.10/m/m.0/attn/MatMul                         BPU  id(0)     HzSQuantizedMatmul         0.992179           6.27205    int8      
/model.10/m/m.0/attn/Mul                            BPU  id(0)     HzSQuantizedConv           0.992179           84.0928    int8      
/model.10/m/m.0/attn/ReduceMax                      BPU  id(0)     HzQuantizedReduceMax       0.997848           14.8656    int8      
/model.10/m/m.0/attn/Sub                            BPU  id(0)     HzSElementwiseSub          0.995229           14.8656    int8      
/model.10/m/m.0/attn/Exp                            BPU  id(0)     HzLut                      0.984391           27.5427    int8      
/model.10/m/m.0/attn/ReduceSum                      BPU  id(0)     HzSQuantizedReduceSum      0.994620           1.0        int8      
/model.10/m/m.0/attn/Div_reciprocal                 BPU  id(0)     HzLut                      0.929943           271.239    int8      
/model.10/m/m.0/attn/Div_mul                        BPU  id(0)     HzSElementwiseMul          0.930553           1.0        int8      
/model.10/m/m.0/attn/Transpose_2                    BPU  id(0)     Transpose                  0.930564           0.339221   int8      
/model.10/m/m.0/attn/MatMul_1                       BPU  id(0)     HzSQuantizedMatmul         0.974196           6.27205    int8      
/model.10/m/m.0/attn/Reshape_1                      BPU  id(0)     Reshape                    0.974196           4.53626    int8      
/model.10/m/m.0/attn/Reshape_2                      BPU  id(0)     Reshape                    0.990036           6.27205    int8      
/model.10/m/m.0/attn/pe/conv/Conv                   BPU  id(0)     HzSQuantizedConv           0.990368           6.27205    int8      
/model.10/m/m.0/attn/proj/conv/Conv                 BPU  id(0)     HzSQuantizedConv           0.939415           3.17294    int8      
/model.10/m/m.0/ffn/ffn.0/conv/Conv                 BPU  id(0)     HzSQuantizedConv           0.985476           6.11307    int8      
/model.10/m/m.0/ffn/ffn.0/act/Mul                   BPU  id(0)     HzLut                      0.961135           5.78014    int8      
/model.10/m/m.0/ffn/ffn.1/conv/Conv                 BPU  id(0)     HzSQuantizedConv           0.985327           3.14796    int8      
/model.10/Split_output_0_calibrated_Requantize      BPU  id(0)     HzRequantize               --                 --         int8      
/model.10/Concat                                    BPU  id(0)     Concat                     0.973990           6.50959    int8      
/model.10/cv2/conv/Conv                             BPU  id(0)     HzSQuantizedConv           0.992625           6.34002    int8      
/model.10/cv2/act/Mul                               BPU  id(0)     HzLut                      0.987348           4.99847    int8      
/model.11/Resize                                    BPU  id(0)     HzQuantizedResizeUpsample  0.987426           4.6676     int8      
/model.11/Resize_output_0_calibrated_Requantize     BPU  id(0)     HzRequantize               --                 --         int8      
...el.6/cv2/act/Mul_output_0_calibrated_Requantize  BPU  id(0)     HzRequantize               --                 --         int8      
/model.12/Concat                                    BPU  id(0)     Concat                     0.985442           4.6676     int8      
/model.13/cv1/conv/Conv                             BPU  id(0)     HzSQuantizedConv           0.993173           3.62884    int8      
/model.13/cv1/act/Mul                               BPU  id(0)     HzLut                      0.988178           5.58117    int8      
/model.13/Split                                     BPU  id(0)     Split                      0.983250           3.65028    int8      
/model.13/m.0/cv1/conv/Conv                         BPU  id(0)     HzSQuantizedConv           0.994485           3.65028    int8      
/model.13/m.0/cv1/act/Mul                           BPU  id(0)     HzLut                      0.990431           5.5279     int8      
/model.13/m.0/cv2/conv/Conv                         BPU  id(0)     HzSQuantizedConv           0.990528           4.57663    int8      
/model.13/m.0/cv2/act/Mul                           BPU  id(0)     HzLut                      0.987291           5.96765    int8      
/model.13/m.0/Add                                   BPU  id(0)     HzSElementwiseAdd          0.988815           3.65028    int8      
/model.13/Split_output_0_calibrated_Requantize      BPU  id(0)     HzRequantize               --                 --         int8      
/model.13/Split_output_1_calibrated_Requantize      BPU  id(0)     HzRequantize               --                 --         int8      
/model.13/Concat                                    BPU  id(0)     Concat                     0.988266           3.65028    int8      
/model.13/cv2/conv/Conv                             BPU  id(0)     HzSQuantizedConv           0.993599           4.47       int8      
/model.13/cv2/act/Mul                               BPU  id(0)     HzLut                      0.989726           4.98534    int8      
/model.14/Resize                                    BPU  id(0)     HzQuantizedResizeUpsample  0.989722           4.58678    int8      
/model.14/Resize_output_0_calibrated_Requantize     BPU  id(0)     HzRequantize               --                 --         int8      
...el.4/cv2/act/Mul_output_0_calibrated_Requantize  BPU  id(0)     HzRequantize               --                 --         int8      
/model.15/Concat                                    BPU  id(0)     Concat                     0.987260           4.58678    int8      
/model.16/cv1/conv/Conv                             BPU  id(0)     HzSQuantizedConv           0.995894           2.9533     int8      
/model.16/cv1/act/Mul                               BPU  id(0)     HzLut                      0.996008           4.92016    int8      
/model.16/Split                                     BPU  id(0)     Split                      0.995574           2.63395    int8      
/model.16/m.0/cv1/conv/Conv                         BPU  id(0)     HzSQuantizedConv           0.993583           2.63395    int8      
/model.16/m.0/cv1/act/Mul                           BPU  id(0)     HzLut                      0.994819           4.47011    int8      
/model.16/m.0/cv2/conv/Conv                         BPU  id(0)     HzSQuantizedConv           0.992435           3.3171     int8      
/model.16/m.0/cv2/act/Mul                           BPU  id(0)     HzLut                      0.992732           5.87229    int8      
/model.16/m.0/Add                                   BPU  id(0)     HzSElementwiseAdd          0.994216           2.63395    int8      
/model.16/Split_output_0_calibrated_Requantize      BPU  id(0)     HzRequantize               --                 --         int8      
/model.16/Split_output_1_calibrated_Requantize      BPU  id(0)     HzRequantize               --                 --         int8      
/model.16/Concat                                    BPU  id(0)     Concat                     0.994941           2.63395    int8      
/model.16/cv2/conv/Conv                             BPU  id(0)     HzSQuantizedConv           0.994310           3.59444    int8      
/model.16/cv2/act/Mul                               BPU  id(0)     HzLut                      0.993924           5.84295    int8      
/model.17/conv/Conv                                 BPU  id(0)     HzSQuantizedConv           0.992467           3.4135     int8      
/model.17/act/Mul                                   BPU  id(0)     HzLut                      0.983971           6.66366    int8      
/model.18/Concat                                    BPU  id(0)     Concat                     0.986936           4.58678    int8      
/model.19/cv1/conv/Conv                             BPU  id(0)     HzSQuantizedConv           0.991467           4.58678    int8      
/model.19/cv1/act/Mul                               BPU  id(0)     HzLut                      0.983050           6.91225    int8      
/model.19/Split                                     BPU  id(0)     Split                      0.978599           5.63601    int8      
/model.19/m.0/cv1/conv/Conv                         BPU  id(0)     HzSQuantizedConv           0.994697           5.63601    int8      
/model.19/m.0/cv1/act/Mul                           BPU  id(0)     HzLut                      0.993075           5.4993     int8      
/model.19/m.0/cv2/conv/Conv                         BPU  id(0)     HzSQuantizedConv           0.988508           3.65437    int8      
/model.19/m.0/cv2/act/Mul                           BPU  id(0)     HzLut                      0.984216           8.63552    int8      
/model.19/m.0/Add                                   BPU  id(0)     HzSElementwiseAdd          0.983302           5.63601    int8      
/model.19/Split_output_0_calibrated_Requantize      BPU  id(0)     HzRequantize               --                 --         int8      
/model.19/Split_output_1_calibrated_Requantize      BPU  id(0)     HzRequantize               --                 --         int8      
/model.19/Concat                                    BPU  id(0)     Concat                     0.982775           5.63601    int8      
/model.19/cv2/conv/Conv                             BPU  id(0)     HzSQuantizedConv           0.989093           6.61649    int8      
/model.19/cv2/act/Mul                               BPU  id(0)     HzLut                      0.982169           9.33676    int8      
/model.20/conv/Conv                                 BPU  id(0)     HzSQuantizedConv           0.990321           6.59445    int8      
/model.20/act/Mul                                   BPU  id(0)     HzLut                      0.982194           6.94996    int8      
/model.21/Concat                                    BPU  id(0)     Concat                     0.985038           4.6676     int8      
/model.22/cv1/conv/Conv                             BPU  id(0)     HzSQuantizedConv           0.988643           4.6676     int8      
/model.22/cv1/act/Mul                               BPU  id(0)     HzLut                      0.978941           7.51227    int8      
/model.22/Split                                     BPU  id(0)     Split                      0.976265           6.80187    int8      
/model.22/m.0/cv1/conv/Conv                         BPU  id(0)     HzSQuantizedConv           0.994550           6.80187    int8      
/model.22/m.0/cv1/act/Mul                           BPU  id(0)     HzLut                      0.994798           4.62989    int8      
/model.22/m.0/m/m.0/cv1/conv/Conv                   BPU  id(0)     HzSQuantizedConv           0.994102           2.43064    int8      
/model.22/m.0/m/m.0/cv1/act/Mul                     BPU  id(0)     HzLut                      0.990746           5.50365    int8      
/model.22/m.0/m/m.0/cv2/conv/Conv                   BPU  id(0)     HzSQuantizedConv           0.995916           4.95399    int8      
/model.22/m.0/m/m.0/cv2/act/Mul                     BPU  id(0)     HzLut                      0.993091           6.10748    int8      
/model.22/m.0/m/m.0/Add                             BPU  id(0)     HzSElementwiseAdd          0.993011           2.43064    int8      
/model.22/m.0/m/m.1/cv1/conv/Conv                   BPU  id(0)     HzSQuantizedConv           0.995823           4.27089    int8      
/model.22/m.0/m/m.1/cv1/act/Mul                     BPU  id(0)     HzLut                      0.991920           6.93543    int8      
/model.22/m.0/m/m.1/cv2/conv/Conv                   BPU  id(0)     HzSQuantizedConv           0.995536           4.67057    int8      
/model.22/m.0/m/m.1/cv2/act/Mul                     BPU  id(0)     HzLut                      0.995182           9.40882    int8      
/model.22/m.0/m/m.1/Add                             BPU  id(0)     HzSElementwiseAdd          0.995131           4.27089    int8      
/model.22/m.0/cv2/conv/Conv                         BPU  id(0)     HzSQuantizedConv           0.984501           6.80187    int8      
/model.22/m.0/cv2/act/Mul                           BPU  id(0)     HzLut                      0.982029           7.08881    int8      
/model.22/m.0/Concat                                BPU  id(0)     Concat                     0.994672           8.89296    int8      
/model.22/m.0/cv3/conv/Conv                         BPU  id(0)     HzSQuantizedConv           0.996007           8.89296    int8      
/model.22/m.0/cv3/act/Mul                           BPU  id(0)     HzLut                      0.993600           7.14319    int8      
/model.22/Split_output_0_calibrated_Requantize      BPU  id(0)     HzRequantize               --                 --         int8      
/model.22/Split_output_1_calibrated_Requantize      BPU  id(0)     HzRequantize               --                 --         int8      
/model.22/Concat                                    BPU  id(0)     Concat                     0.988210           6.80187    int8      
/model.22/cv2/conv/Conv                             BPU  id(0)     HzSQuantizedConv           0.993425           6.33787    int8      
/model.22/cv2/act/Mul                               BPU  id(0)     HzLut                      0.988929           11.0639    int8      
/model.23/cv3.0/cv3.0.0/cv3.0.0.0/conv/Conv         BPU  id(0)     HzSQuantizedConv           0.998233           3.4135     int8      
/model.23/cv3.0/cv3.0.0/cv3.0.0.0/act/Mul           BPU  id(0)     HzLut                      0.997973           6.01443    int8      
/model.23/cv3.0/cv3.0.0/cv3.0.0.1/conv/Conv         BPU  id(0)     HzSQuantizedConv           0.989426           5.87941    int8      
/model.23/cv3.0/cv3.0.0/cv3.0.0.1/act/Mul           BPU  id(0)     HzLut                      0.986380           7.82783    int8      
/model.23/cv3.0/cv3.0.1/cv3.0.1.0/conv/Conv         BPU  id(0)     HzSQuantizedConv           0.990949           5.86644    int8      
/model.23/cv3.0/cv3.0.1/cv3.0.1.0/act/Mul           BPU  id(0)     HzLut                      0.992433           10.2571    int8      
/model.23/cv3.0/cv3.0.1/cv3.0.1.1/conv/Conv         BPU  id(0)     HzSQuantizedConv           0.990006           7.49609    int8      
/model.23/cv3.0/cv3.0.1/cv3.0.1.1/act/Mul           BPU  id(0)     HzLut                      0.991923           28.9628    int8      
/model.23/cv3.0/cv3.0.2/Conv                        BPU  id(0)     HzSQuantizedConv           0.999801           26.4145    int8      
/model.23/cv2.0/cv2.0.0/conv/Conv                   BPU  id(0)     HzSQuantizedConv           0.990723           3.4135     int8      
/model.23/cv2.0/cv2.0.0/act/Mul                     BPU  id(0)     HzLut                      0.988540           12.1493    int8      
/model.23/cv2.0/cv2.0.1/conv/Conv                   BPU  id(0)     HzSQuantizedConv           0.986628           7.68527    int8      
/model.23/cv2.0/cv2.0.1/act/Mul                     BPU  id(0)     HzLut                      0.988132           27.8172    int8      
/model.23/cv2.0/cv2.0.2/Conv                        BPU  id(0)     HzSQuantizedConv           0.995825           20.4675    int8      
/model.23/cv3.1/cv3.1.0/cv3.1.0.0/conv/Conv         BPU  id(0)     HzSQuantizedConv           0.990146           6.59445    int8      
/model.23/cv3.1/cv3.1.0/cv3.1.0.0/act/Mul           BPU  id(0)     HzLut                      0.989963           7.09282    int8      
/model.23/cv3.1/cv3.1.0/cv3.1.0.1/conv/Conv         BPU  id(0)     HzSQuantizedConv           0.988528           5.51392    int8      
/model.23/cv3.1/cv3.1.0/cv3.1.0.1/act/Mul           BPU  id(0)     HzLut                      0.984898           6.76259    int8      
/model.23/cv3.1/cv3.1.1/cv3.1.1.0/conv/Conv         BPU  id(0)     HzSQuantizedConv           0.993969           5.68406    int8      
/model.23/cv3.1/cv3.1.1/cv3.1.1.0/act/Mul           BPU  id(0)     HzLut                      0.993995           7.63307    int8      
/model.23/cv3.1/cv3.1.1/cv3.1.1.1/conv/Conv         BPU  id(0)     HzSQuantizedConv           0.990762           6.52937    int8      
/model.23/cv3.1/cv3.1.1/cv3.1.1.1/act/Mul           BPU  id(0)     HzLut                      0.991157           36.0001    int8      
/model.23/cv3.1/cv3.1.2/Conv                        BPU  id(0)     HzSQuantizedConv           0.999840           32.4379    int8      
/model.23/cv2.1/cv2.1.0/conv/Conv                   BPU  id(0)     HzSQuantizedConv           0.989680           6.59445    int8      
/model.23/cv2.1/cv2.1.0/act/Mul                     BPU  id(0)     HzLut                      0.988603           10.9617    int8      
/model.23/cv2.1/cv2.1.1/conv/Conv                   BPU  id(0)     HzSQuantizedConv           0.987141           7.53092    int8      
/model.23/cv2.1/cv2.1.1/act/Mul                     BPU  id(0)     HzLut                      0.988374           17.895     int8      
/model.23/cv2.1/cv2.1.2/Conv                        BPU  id(0)     HzSQuantizedConv           0.995120           16.3473    int8      
/model.23/cv3.2/cv3.2.0/cv3.2.0.0/conv/Conv         BPU  id(0)     HzSQuantizedConv           0.991266           10.8162    int8      
/model.23/cv3.2/cv3.2.0/cv3.2.0.0/act/Mul           BPU  id(0)     HzLut                      0.991072           5.99436    int8      
/model.23/cv3.2/cv3.2.0/cv3.2.0.1/conv/Conv         BPU  id(0)     HzSQuantizedConv           0.997060           4.83003    int8      
/model.23/cv3.2/cv3.2.0/cv3.2.0.1/act/Mul           BPU  id(0)     HzLut                      0.996911           5.42119    int8      
/model.23/cv3.2/cv3.2.1/cv3.2.1.0/conv/Conv         BPU  id(0)     HzSQuantizedConv           0.997902           4.70067    int8      
/model.23/cv3.2/cv3.2.1/cv3.2.1.0/act/Mul           BPU  id(0)     HzLut                      0.998482           3.61351    int8      
/model.23/cv3.2/cv3.2.1/cv3.2.1.1/conv/Conv         BPU  id(0)     HzSQuantizedConv           0.998193           2.78216    int8      
/model.23/cv3.2/cv3.2.1/cv3.2.1.1/act/Mul           BPU  id(0)     HzLut                      0.998506           10.4103    int8      
/model.23/cv3.2/cv3.2.2/Conv                        BPU  id(0)     HzSQuantizedConv           0.999971           9.98083    int8      
/model.23/cv2.2/cv2.2.0/conv/Conv                   BPU  id(0)     HzSQuantizedConv           0.997146           10.8162    int8      
/model.23/cv2.2/cv2.2.0/act/Mul                     BPU  id(0)     HzLut                      0.995881           8.56178    int8      
/model.23/cv2.2/cv2.2.1/conv/Conv                   BPU  id(0)     HzSQuantizedConv           0.997379           8.56014    int8      
/model.23/cv2.2/cv2.2.1/act/Mul                     BPU  id(0)     HzLut                      0.997570           11.2234    int8      
/model.23/cv2.2/cv2.2.2/Conv                        BPU  id(0)     HzSQuantizedConv           0.997736           11.2233    int8
2025-09-14 19:19:09,926 INFO The quantized model output:
=============================================================================
Output      Cosine Similarity  L1 Distance  L2 Distance  Chebyshev Distance  
-----------------------------------------------------------------------------
output0     0.999801           0.183934     0.001493     7.302748            
480         0.995825           0.134839     0.000321     4.977159            
494         0.999840           0.163098     0.002707     4.029671            
502         0.995120           0.151466     0.000710     3.849218            
516         0.999971           0.059118     0.001974     1.133204            
524         0.997736           0.188838     0.001714     1.374329
2025-09-14 19:19:09,931 INFO End to Horizon NN Model Convert.
2025-09-14 19:19:09,946 INFO start convert to *.bin file....
2025-09-14 19:19:09,980 INFO ONNX model output num : 6
2025-09-14 19:19:09,984 INFO ############# model deps info #############
2025-09-14 19:19:09,984 INFO hb_mapper version   : 1.24.3
2025-09-14 19:19:09,984 INFO hbdk version        : 3.49.15
2025-09-14 19:19:09,984 INFO hbdk runtime version: 3.15.55.0
2025-09-14 19:19:09,984 INFO horizon_nn version  : 1.1.0
2025-09-14 19:19:09,984 INFO ############# model_parameters info #############
2025-09-14 19:19:09,984 INFO onnx_model          : /open_explorer/best.onnx
2025-09-14 19:19:09,984 INFO BPU march           : bayes-e
2025-09-14 19:19:09,984 INFO layer_out_dump      : False
2025-09-14 19:19:09,984 INFO log_level           : DEBUG
2025-09-14 19:19:09,984 INFO working dir         : /open_explorer/.temporary_workspace/bpu_model_output
2025-09-14 19:19:09,985 INFO output_model_file_prefix: best_bayese_640x640_nv12
2025-09-14 19:19:09,985 INFO ############# input_parameters info #############
2025-09-14 19:19:09,985 INFO ------------------------------------------
2025-09-14 19:19:09,985 INFO ---------input info : images ---------
2025-09-14 19:19:09,985 INFO input_name          : images
2025-09-14 19:19:09,985 INFO input_type_rt       : nv12
2025-09-14 19:19:09,985 INFO input_space&range   : regular
2025-09-14 19:19:09,985 INFO input_layout_rt     : None
2025-09-14 19:19:09,985 INFO input_type_train    : rgb
2025-09-14 19:19:09,985 INFO input_layout_train  : NCHW
2025-09-14 19:19:09,985 INFO norm_type           : data_scale
2025-09-14 19:19:09,985 INFO input_shape         : 1x3x640x640
2025-09-14 19:19:09,985 INFO scale_value         : 0.003921568627451,
2025-09-14 19:19:09,985 INFO cal_data_dir        : /open_explorer/.temporary_workspace/.calibration_data_temporary_folder
2025-09-14 19:19:09,985 INFO cal_data_type       : float32
2025-09-14 19:19:09,985 INFO ---------input info : images end -------
2025-09-14 19:19:09,986 INFO ------------------------------------------
2025-09-14 19:19:09,986 INFO ############# calibration_parameters info #############
2025-09-14 19:19:09,986 INFO preprocess_on       : False
2025-09-14 19:19:09,986 INFO calibration_type:   : default
2025-09-14 19:19:09,986 INFO optimization        : set_Softmax_input_int8,set_Softmax_output_int8;
2025-09-14 19:19:09,986 INFO per_channel         : False
2025-09-14 19:19:09,986 INFO ############# compiler_parameters info #############
2025-09-14 19:19:09,986 INFO debug               : True
2025-09-14 19:19:09,986 INFO compile_mode        : latency
2025-09-14 19:19:09,986 INFO hbdk_pass_through_params: --O3 --debug --core-num 1 --fast --jobs 16
2025-09-14 19:19:09,986 INFO input-source        : {'images': 'pyramid', '_default_value': 'ddr'}
2025-09-14 19:19:09,995 INFO Convert to runtime bin file successfully!
2025-09-14 19:19:09,995 INFO End Model Convert
[MZOO] [19:19:10.516] [INFO] Model conversion completed successfully!
[MZOO] [19:19:10.516] [INFO] Looking for output file: /open_explorer/.temporary_workspace/bpu_model_output/best_bayese_640x640_nv12.bin
[MZOO] [19:19:10.516] [INFO] Output file moved to: /open_explorer/best_bayese_640x640_nv12.bin
[MZOO] [19:19:10.516] [INFO] hb_mapper log moved to: /open_explorer/hb_mapper_makertbin.log
[MZOO] [19:19:10.516] [INFO] Restored working directory to: /open_explorer
[MZOO] [19:19:10.516] [INFO] Cleaning up...
[MZOO] [19:19:10.563] [INFO] Temporary files cleaned up.
[MZOO] [19:19:10.563] [INFO] Conversion completed successfully!
[MZOO] [19:19:10.563] [INFO] Output file: /open_explorer/best_bayese_640x640_nv12.bin
[MZOO] [19:19:10.563] [INFO] Mapper log: /open_explorer/hb_mapper_makertbin.log

if you want reproduce this results, please run the following bash command.

bash is here

export CAM_TYPE=usb
source /opt/tros/humble/setup.bash
ros2 launch dnn_node_example dnn_node_example.launch.py dnn_example_config_file:=yolov11workconfig-rm.json dnn_example_width:=640 dnn_example_width:=640

yolov11workconfig-rm.json is here

{
	"model_file": "/home/sunrise/robomaster_damagepanel/best_bayese_640x640_nv12.bin",
	"task_num": 4,
	"dnn_Parser": "yolov8",
	"model_output_count": 6,
	"reg_max": 16,
	"class_num": 5,
	"cls_names_list": "config/rm.list",
	"strides": [8, 16, 32],
	"score_threshold": 0.25,
	"nms_threshold": 0.7,
	"nms_top_k": 300,
	"output_order": [0,1,2,3,4,5]
}

config/rm.list is here

panel
base
robot
ignore
sentry

this is a log of ros2 launch dnn_node_example dnn_node_example.launch.py dnn_example_config_file:=yolov11workconfig-rm.json dnn_example_width:=640 dnn_example_width:=640 

sunrise@ubuntu:~/dnn_node_example_config$ ros2 launch dnn_node_example dnn_node_example.launch.py dnn_example_config_file:=/home/sunrise/dnn_node_example_config/yolov11workconfig-rm.json dnn_example_width:=640 dnn_example_width:=640
[INFO] [launch]: All log files can be found below /home/sunrise/.ros/log/2025-09-14-20-24-23-506205-ubuntu-101852
[INFO] [launch]: Default logging verbosity is set to INFO
dnn_node_example_path is  /opt/tros/humble/lib/dnn_node_example
cp_cmd is  cp -r /opt/tros/humble/lib/dnn_node_example/config .
camera_type is  usb
using usb cam
Hobot shm pkg enables zero-copy with fastrtps profiles file: /opt/tros/humble/lib/hobot_shm/config/shm_fastdds.xml
Hobot shm pkg sets RMW_FASTRTPS_USE_QOS_FROM_XML: 1
config_file_path is  /opt/tros/humble/lib/hobot_usb_cam/config/usb_camera_calibration.yaml
env of RMW_FASTRTPS_USE_QOS_FROM_XML is  1 , ignore env setting
env of RMW_FASTRTPS_USE_QOS_FROM_XML is  1 , ignore env setting
webserver has launch
env of RMW_FASTRTPS_USE_QOS_FROM_XML is  1 , ignore env setting
config_file_path is  /opt/tros/humble/lib/hobot_usb_cam/config/usb_camera_calibration.yaml
env of RMW_FASTRTPS_USE_QOS_FROM_XML is  1 , ignore env setting
env of RMW_FASTRTPS_USE_QOS_FROM_XML is  1 , ignore env setting
env of RMW_FASTRTPS_USE_QOS_FROM_XML is  1 , ignore env setting
env of RMW_FASTRTPS_USE_QOS_FROM_XML is  1 , ignore env setting
webserver has launch
[INFO] [hobot_usb_cam-1]: process started with pid [101858]
[INFO] [hobot_codec_republish-2]: process started with pid [101860]
[INFO] [example-3]: process started with pid [101862]
[INFO] [websocket-4]: process started with pid [101864]
[hobot_codec_republish-2] [WARN] [1757852664.136454163] [hobot_codec_decoder]: Parameters:
[hobot_codec_republish-2] sub_topic: /image
[hobot_codec_republish-2] pub_topic: /hbmem_img
[hobot_codec_republish-2] channel: 1
[hobot_codec_republish-2] in_mode: ros
[hobot_codec_republish-2] out_mode: shared_mem
[hobot_codec_republish-2] in_format: jpeg
[hobot_codec_republish-2] out_format: nv12
[hobot_codec_republish-2] enc_qp: 10
[hobot_codec_republish-2] jpg_quality: 60
[hobot_codec_republish-2] input_framerate: 30
[hobot_codec_republish-2] output_framerate: -1
[hobot_codec_republish-2] dump_output: 0
[hobot_codec_republish-2] [WARN] [1757852664.148894399] [HobotCodecImpl]: platform x5
[hobot_usb_cam-1] [WARN] [1757852664.297457687] [hobot_usb_cam]: framerate: 30
[hobot_usb_cam-1] [WARN] [1757852664.298013228] [hobot_usb_cam]: pixel_format_name: mjpeg
[hobot_usb_cam-1] [WARN] [1757852664.321911034] [hobot_usb_cam]: Camera calibration file: [/opt/tros/humble/lib/hobot_usb_cam/config/usb_camera_calibration.yaml] does not exist!
[hobot_usb_cam-1] If you need calibration msg, please make sure the calibration file path is correct and the calibration file exists!
[hobot_usb_cam-1] [WARN] [1757852664.333354271] [hobot_usb_cam]: Try to open device [/dev/video0]
[hobot_usb_cam-1] [WARN] [1757852664.423365835] [hobot_usb_cam]: This devices supproted formats:
[hobot_usb_cam-1] [WARN] [1757852664.423569252] [hobot_usb_cam]: 	Motion-JPEG: 1920 x 1080 (30 Hz)
[hobot_usb_cam-1] [WARN] [1757852664.423632377] [hobot_usb_cam]: 	Motion-JPEG: 1280 x 960 (30 Hz)
[hobot_usb_cam-1] [WARN] [1757852664.423674043] [hobot_usb_cam]: 	Motion-JPEG: 1280 x 720 (30 Hz)
[hobot_usb_cam-1] [WARN] [1757852664.423713127] [hobot_usb_cam]: 	Motion-JPEG: 1024 x 576 (30 Hz)
[hobot_usb_cam-1] [WARN] [1757852664.423751168] [hobot_usb_cam]: 	Motion-JPEG: 800 x 600 (30 Hz)
[hobot_usb_cam-1] [WARN] [1757852664.423789001] [hobot_usb_cam]: 	Motion-JPEG: 640 x 480 (30 Hz)
[hobot_usb_cam-1] [WARN] [1757852664.423825960] [hobot_usb_cam]: 	Motion-JPEG: 640 x 360 (30 Hz)
[hobot_usb_cam-1] [WARN] [1757852664.423863251] [hobot_usb_cam]: 	YUYV 4:2:2: 640 x 480 (30 Hz)
[hobot_usb_cam-1] [WARN] [1757852664.423900293] [hobot_usb_cam]: 	YUYV 4:2:2: 640 x 360 (30 Hz)
[hobot_usb_cam-1] [WARN] [1757852664.431134701] [hobot_usb_cam]: Open & Init device /dev/video0 success.
[example-3] [BPU_PLAT]BPU Platform Version(1.3.6)!
[example-3] [HBRT] set log level as 0. version = 3.15.55.0
[example-3] [DNN] Runtime version = 1.24.5_(3.15.55 HBRT)
[websocket-4] [WARN] [1757852664.501137205] [websocket]: 
[websocket-4] Parameter:
[websocket-4]  image_topic: /image
[websocket-4]  image_type: mjpeg
[websocket-4]  only_show_image: 0
[websocket-4]  smart_topic: hobot_dnn_detection
[websocket-4]  output_fps: 0
[example-3] [A][DNN][packed_model.cpp:247][Model](2025-09-14,20:24:24.682.335) [HorizonRT] The model builder version = 1.24.3
[ERROR] [example-3]: process has died [pid 101862, exit code -11, cmd '/opt/tros/humble/lib/dnn_node_example/example --ros-args --log-level error --ros-args --params-file /tmp/launch_params_db44mrdw --params-file /tmp/launch_params_puvuqml1 --params-file /tmp/launch_params_20z6hjbl --params-file /tmp/launch_params_97cdcycp --params-file /tmp/launch_params_44ig_tci'].
[websocket-4] [WARN] [1757852665.604772611] [websocket]: web socket has cache image num > 20, size 21
[websocket-4] [WARN] [1757852665.636796117] [websocket]: web socket has cache image num > 20, size 22
[websocket-4] [WARN] [1757852665.672856367] [websocket]: web socket has cache image num > 20, size 23
[websocket-4] [WARN] [1757852665.704739247] [websocket]: web socket has cache image num > 20, size 24
[websocket-4] [WARN] [1757852665.740689581] [websocket]: web socket has cache image num > 20, size 25
[websocket-4] [WARN] [1757852665.772596878] [websocket]: web socket has cache image num > 20, size 26
[websocket-4] [WARN] [1757852665.808784503] [websocket]: web socket has cache image num > 20, size 27
[websocket-4] [WARN] [1757852665.840743134] [websocket]: web socket has cache image num > 20, size 28
[websocket-4] [WARN] [1757852665.876704551] [websocket]: web socket has cache image num > 20, size 29
[websocket-4] [WARN] [1757852665.908763973] [websocket]: web socket has cache image num > 20, size 30
[websocket-4] [WARN] [1757852665.944663057] [websocket]: web socket has cache image num > 20, size 31
[websocket-4] [WARN] [1757852665.976770728] [websocket]: web socket has cache image num > 20, size 32
[websocket-4] [WARN] [1757852666.012892020] [websocket]: web socket has cache image num > 20, size 33
[websocket-4] [WARN] [1757852666.044731359] [websocket]: web socket has cache image num > 20, size 34
[websocket-4] [WARN] [1757852666.079789444] [websocket]: web socket has cache image num > 20, size 35
[websocket-4] [WARN] [1757852666.112889240] [websocket]: web socket has cache image num > 20, size 36
[websocket-4] [WARN] [1757852666.148663115] [websocket]: web socket has cache image num > 20, size 37
[websocket-4] [WARN] [1757852666.180728454] [websocket]: web socket has cache image num > 20, size 38
[websocket-4] [WARN] [1757852666.212784042] [websocket]: web socket has cache image num > 20, size 39
[websocket-4] [WARN] [1757852666.248659001] [websocket]: web socket has cache image num > 20, size 40
[websocket-4] [WARN] [1757852666.280664590] [websocket]: web socket has cache image num > 20, size 41
[websocket-4] [WARN] [1757852666.316696090] [websocket]: web socket has cache image num > 20, size 42
[websocket-4] [WARN] [1757852666.348685804] [websocket]: web socket has cache image num > 20, size 43
[websocket-4] [WARN] [1757852666.384645138] [websocket]: web socket has cache image num > 20, size 44
[websocket-4] [WARN] [1757852666.415472436] [websocket]: web socket has cache image num > 20, size 45
[websocket-4] [WARN] [1757852666.452770101] [websocket]: web socket has cache image num > 20, size 46
[websocket-4] [WARN] [1757852666.484664107] [websocket]: web socket has cache image num > 20, size 47
[websocket-4] [WARN] [1757852666.520685774] [websocket]: web socket has cache image num > 20, size 48
[websocket-4] [WARN] [1757852666.551363697] [websocket]: web socket has cache image num > 20, size 49
[websocket-4] [WARN] [1757852666.588616030] [websocket]: web socket has cache image num > 20, size 50
[websocket-4] [WARN] [1757852666.620660993] [websocket]: web socket has cache image num > 20, size 51
[websocket-4] [WARN] [1757852666.656631369] [websocket]: web socket has cache image num > 20, size 52
[websocket-4] [WARN] [1757852666.688654124] [websocket]: web socket has cache image num > 20, size 53
[websocket-4] [WARN] [1757852666.724671374] [websocket]: web socket has cache image num > 20, size 54
[websocket-4] [WARN] [1757852666.756620421] [websocket]: web socket has cache image num > 20, size 55
[websocket-4] [WARN] [1757852666.792703422] [websocket]: web socket has cache image num > 20, size 56
[websocket-4] [WARN] [1757852666.824621969] [websocket]: web socket has cache image num > 20, size 57
[websocket-4] [WARN] [1757852666.856677182] [websocket]: web socket has cache image num > 20, size 58
[websocket-4] [WARN] [1757852666.892675808] [websocket]: web socket has cache image num > 20, size 59
[websocket-4] [WARN] [1757852666.924691938] [websocket]: web socket has cache image num > 20, size 60
[websocket-4] [WARN] [1757852666.960593438] [websocket]: web socket has cache image num > 20, size 61
^C[WARNING] [launch]: user interrupted with ctrl-c (SIGINT)
[websocket-4] [WARN] [1757852666.976859753] [websocket]: web socket has cache image num > 20, size 62
[hobot_usb_cam-1] [WARN] [1757852666.991104737] [hobot_usb_cam]: Shutting down
[INFO] [hobot_codec_republish-2]: process has finished cleanly [pid 101860]
[INFO] [websocket-4]: process has finished cleanly [pid 101864]
[INFO] [hobot_usb_cam-1]: process has finished cleanly [pid 101858]

Thanks again for your support!