yingkunwu / R-YOLOv4

This is a PyTorch-based R-YOLOv4 implementation which combines YOLOv4 model and loss function from R3Det for arbitrary oriented object detection.
114 stars 20 forks source link

tensor wrong size #69

Open mahyredaa opened 1 week ago

mahyredaa commented 1 week ago

I'm trying to implement the YOLOv8 and YOLOv9 on the code as other versions, but there is always a problem with the reshaping of the output tensor from the YOLO model, the outputs i get is torch.Size([4, 512, 100, 100]), while the output that should be is torch.Size([4, 603, 100, 100]), though I'm implementing the YOLO correctly.