openvino object-detection yoloV5 architecture

[TusharWagh]

I am trying object detection on custom/self trained yoloV5 model. I am referring https://github.com/ultralytics/yolov5.
After training this model, I exported weights into onnx format. Then I converted onnx model into openvino IR. 
When tried openvino object-detection, I got error.
This is error :

D:\projects\openvino_yoloV5>python object_detection_demo.py --model model\best.xml --input input\test_image.jpg --architecture_type yolo

I attached model files in attachments section.

Error:
[ INFO ] Initializing Inference Engine...
[ INFO ] Loading network...
[ INFO ] Reading network from IR...
Traceback (most recent call last):
File "object_detection_demo.py", line 326, in <module>
sys.exit(main() or 0)
File "object_detection_demo.py", line 218, in main
model = get_model(ie, args)
File "object_detection_demo.py", line 132, in get_model
threshold=args.prob_threshold, keep_aspect_ratio=args.keep_aspect_ratio)
File "D:\projects\openvino_yoloV5\common\models\yolo.py", line 72, in __init__
self.yolo_layer_params = self._get_output_info()
File "D:\projects\openvino_yoloV5\common\models\yolo.py", line 83, in _get_output_info
shape = list(get_parent(node).shape)
ValueError: get_shape was called on a descriptor::Tensor with dynamic shape

[Zulkifli_Intel · Accepted solution]

Hello Tushar Wagh,

Thank you for contacting us.

YOLOv5 is currently not an officially supported topology by the OpenVINO toolkit nor has it been validated by us. It will be supported in future versions of the OpenVINO Toolkit.

Regards,

Zulkifli

[Zulkifli_Intel]

Hello Tushar Wagh,

This thread will no longer be monitored since this issue has been resolved. If you need any additional information from Intel, please submit a new question.

Regards,

Zulkifli