Sai Krishna Das

I got it working after a lot of debugging inside my code. In my case, the permission problem was because of the way I opened the image `img = cv2.imread('G:/project/OCR/FDA.png')`...

converting darknet to yolov4 works fine. But its too slow to run on my jetson nx (took about 64 secs for detection ). I need to convert the yolov4.pt file...

Doesn't that convert yolov4.weight to yolov4.wts and to .engine? I have weight named yolov4.pt

@wang-xinyu NameError : name 'device' is not defined. in model.load_state_dict(torch.load(weights,map_location=device)['model'])

But my training process is exited after showing this.

Do you have a solution for this? This usually happens when I use more than one GPU with p7 model @WongKinYiu

Check this screenshot, I re-installed tensorboard. But now i get a cuda out of memory.  @WongKinYiu

My original image size is 1280 x 720 (width 1280 and height 720 pixels). Which input size is good to use if i need to train on a p7 model...

@WongKinYiu

`python -m torch.distributed.launch --nproc_per_node 4 train.py --batch-size 16 --img 1536 1536 --data '../data.yaml' --cfg ./models/yolov4-p7.yaml --weights '' --sync-bn --device 0,1,2,3 --name yolov4-p7` I used this cmd to train the model....