rknn_model_zoo 模型转换失败

• Hugh

  Posts: 2

  Joined: 2024-11-21 14:52

  2024-11-21 15:02

  我购买了Luckfox-Pico-Max，并烧录了网盘提供的Buildroot FLASH版本系统。但问题与此无关。
  在尝试复现RKNN 推理测试_6.1 导出 RKNN 模型中，执行 rknn_model_zoo/examples/yolov5/python 目录下的模型转换程序 convert.py报错。请问如何解决？
  环境为自建虚拟机：ubuntu-22.04.5，RKNN-Toolkit2版本为1.60（原仓库已声明废止，最新版本为2.30，是否应该升级），rknn_model_zoo为最新2.30。

  Code: Select all

  (RKNN-Toolkit2) ubuntu@ubuntu:~/Projects/rknn_model_zoo/examples/yolov5/python$ python3 convert.py ../model/yolov5s.onnx rv1106
  W __init__: rknn-toolkit2 version: 1.6.0+81f21f4d
  --> Config model
  done
  --> Loading model
  E load_onnx: Model file ../model/yolov5s.onnx not exists!
  W load_onnx: ===================== WARN(1) =====================
  E rknn-toolkit2 version: 1.6.0+81f21f4d
  Traceback (most recent call last):
    File "convert.py", line 52, in <module>
      ret = rknn.load_onnx(model=model_path)
    File "/home/ubuntu/miniconda3/envs/RKNN-Toolkit2/lib/python3.8/site-packages/rknn/api/rknn.py", line 168, in load_onnx
      return self.rknn_base.load_onnx(model, inputs, input_size_list, input_initial_val, outputs)
    File "rknn/api/rknn_base.py", line 1441, in rknn.api.rknn_base.RKNNBase.load_onnx
    File "rknn/api/rknn_log.py", line 92, in rknn.api.rknn_log.RKNNLog.e
  ValueError: Model file ../model/yolov5s.onnx not exists!
  (RKNN-Toolkit2) ubuntu@ubuntu:~/Projects/rknn_model_zoo/examples/yolov5/python$ python3 convert.py ../model/yolov5s_relu.onnx rv1106
  W __init__: rknn-toolkit2 version: 1.6.0+81f21f4d
  --> Config model
  done
  --> Loading model
  W load_onnx: It is recommended onnx opset 19, but your onnx model opset is 12!
  W load_onnx: Model converted from pytorch, 'opset_version' should be set 19 in torch.onnx.export for successful convert!
  Loading :   0%|                                                             | 0/125 [00:00<?, ?Loading : 100%|█████████████████████████████████████████████████| 125/125 [00:00<00:00, 3534.75it/s]
  done
  --> Building model
  Illegal instruction (core dumped)

  • Quote

(RKNN-Toolkit2) ubuntu@ubuntu:~/Projects/rknn_model_zoo/examples/yolov5/python$ python3 convert.py ../model/yolov5s_relu.onnx rv1106
W __init__: rknn-toolkit2 version: 1.6.0+81f21f4d
--> Config model
done
--> Loading model
W load_onnx: It is recommended onnx opset 19, but your onnx model opset is 12!
W load_onnx: Model converted from pytorch, 'opset_version' should be set 19 in torch.onnx.export for successful convert!
Loading :   0%|                                                             | 0/125 [00:00<?, ?Loading : 100%|█████████████████████████████████████████████████| 125/125 [00:00<00:00, 3534.75it/s]
done
--> Building model
Illegal instruction (core dumped)

(RKNN-Toolkit2) ➜  pythonpython3 convert.py ../model/yolov5s_relu.onnx rv1106
I rknn-toolkit2 version: 2.0.0b0+9bab5682
--> Config model
done
--> Loading model
I It is recommended onnx opset 19, but your onnx model opset is 12!
I Model converted from pytorch, 'opset_version' should be set 19 in torch.onnx.export for successful convert!
I Loading : 100%|██████████████████████████████████████████████| 125/125 [00:00<00:00, 14088.46it/s]
done
--> Building model
I GraphPreparing : 100%|████████████████████████████████████████| 149/149 [00:00<00:00, 8050.07it/s]
I Quantizating : 100%|████████████████████████████████████████████| 149/149 [00:04<00:00, 30.34it/s]
W build: The default input dtype of 'images' is changed from 'float32' to 'int8' in rknn model for performance!
                       Please take care of this change when deploy rknn model with Runtime API!
W build: The default output dtype of 'output0' is changed from 'float32' to 'int8' in rknn model for performance!
                      Please take care of this change when deploy rknn model with Runtime API!
W build: The default output dtype of '286' is changed from 'float32' to 'int8' in rknn model for performance!
                      Please take care of this change when deploy rknn model with Runtime API!
W build: The default output dtype of '288' is changed from 'float32' to 'int8' in rknn model for performance!
                      Please take care of this change when deploy rknn model with Runtime API!
I rknn building ...
I rknn buiding done.
done
--> Export rknn model
done