Closed ltetrel closed 5 days ago
Hi all,
At Kitware, we worked on converting a darknet yolo model to ONNX. Here is the link to the public repo for anyone who is interrested: https://gitlab.kitware.com/keu-computervision/ml/darknet-to-pytorch-onnx It is based from Tianxiaomo work but updated for latest yolov4. The ideal goal is to use https://github.com/jkjung-avt/tensorrt_demos/blob/master/yolo/yolo_to_onnx.py for the "base" model, then add the output layer (bounding boxes extraction from conv layers) only in evaluation mode.
It will be soon integrated in VIAME.
Hi all,
At Kitware, we worked on converting a darknet yolo model to ONNX. Here is the link to the public repo for anyone who is interrested: https://gitlab.kitware.com/keu-computervision/ml/darknet-to-pytorch-onnx It is based from Tianxiaomo work but updated for latest yolov4. The ideal goal is to use https://github.com/jkjung-avt/tensorrt_demos/blob/master/yolo/yolo_to_onnx.py for the "base" model, then add the output layer (bounding boxes extraction from conv layers) only in evaluation mode.
It will be soon integrated in VIAME.