来自于USTC的2022秋季数字图像处理project
本课程设计基于cuda 11.7和YOLO v7,引入了Swin transformer V2和BOTNET,COTNET,做了比较~
一,RuntimeError: Expected all tensors to be on the same device, but found at least two devices, cuda:0 and cpu!
modify:在报错的地方定义device
device = torch.device('cuda:0')
issue:目前Swin transformer V2还不能这样修改,我是在CPU上跑得~