Ayush Chaurasia
Ayush Chaurasia
No it's not. Could you describe your use case with darknet?
You can train any model with ultralytics repo. You can even train yolov4,v3 or even non-yolo model. You just have to prepare the cfg file properly. We'll add support for...
resume might not be working as expected. This is a known issue along with tf export as mentioned on the readme. We'll fix this soon
results is a list. see example here https://docs.ultralytics.com/predict/
Are you talking about 3d detection?
@Laughing-q this should be possible right?
@dmddmd the boxes class is a wrapper around tensor. you can access the raw tensor directly by doing boxes.data..
It might be related to your opencv package. Is the `opencv-contrib-python`same as `pip install cv2-python` ?
We'll work on supporting it soon
The last val is run independently after training is finished. The val=False was intended for "turning off validation during training". But I agree that this might be confusing to some...