IndexError: invalid index of a 0-dim tensor

[zjykzj]

Met this question when using predict.py. Error code location is located at

        i = order[0]
        keep.append(i)

after check the code, I found the key is to update squeeze() use

        # before
        ids = (ovr<=threshold).nonzero().squeeze()
        # after
        ids = (ovr<=threshold).nonzero().squeeze(-1)

this change can solve the problem, when ids shape == (1,1), then squeeze(-1) return (1)

[AlbertMa123]

应该是pytorch版本升级导致的问题，最后一次order，order=tensor(25)（例如），此时shape=[]，即不能用order[0]来获取第一个元素，建议这么写：
if order.dim()==0:
order = order.reshape(-1)

[zjykzj]

应该是pytorch版本升级导致的问题，最后一次order，order=tensor(25)（例如），此时shape=[]，即不能用order[0]来获取第一个元素，建议这么写： if order.dim()==0: order = order.reshape(-1)

Yes! YOLOv1 is a product of about 10 years ago, so there will be many changes in the current software environment. I think most people will pay attention to the open-source works that have implemented algorithms before in order to better delve into the field of computer vision. I have also tried to implement some YOLO series algorithms, and those interested can check them out:

• zjykzj/YOLO5Face
• zjykzj/YOLOv3
• zjykzj/YOLOv2
• zjykzj/YOLOv1