sky

请问执行train.py时 ，dataset.py文件的52行 img值为什么是NONE，buf是正常的 @youngyoung611

@haritha91 excuse me, In /mmaction/datasets/pipelines/loading.py `@PIPELINES.register_module() class LoadLocalizationFeature: def __init__(self, raw_feature_ext='.csv'): **valid_raw_feature_ext = ('.csv', )** if raw_feature_ext not in valid_raw_feature_ext: raise NotImplementedError self.raw_feature_ext = raw_feature_ext ` What is the content...

@ersheng-ai thanks your share, but QR code expired.Can you Update?

@kraten Thanks for sharing, but I think there are some bugs in the above code: due to the influence of perspective deformation, the distance between pixels is not always linear....

@wanghao9610 Have you found the relevant code?

> ---> 32 model.load_state_dict(checkpoint['state_dict']) > 33 # model.load_state_dict(torch.load("C:\Users####\pytorch-video-recognition\c3d-pretrained.pth.tar", map_location=lambda storage, loc: storage)) > 34 > > KeyError: 'state_dict' model = C3D.C3D(num_classes=101, pretrained=True) # checkpoint = torch.load('c3d-pretrained.pth', map_location=lambda storage, loc: storage)...

> Can you run inference.py with pretrained model? > I run inference.py and see error: > Traceback (most recent call last): > File "inference.py", line 78, in > main() >...

> I didn't solve it in the end, even though I asked for help via email. > […](#) > ------------------ 原始邮件 ------------------ 发件人: "jfzhang95/pytorch-video-recognition"

@jfzhang95 I meet the same error, when I train from scratch on ucf101. The accuracy is very low about(0.001). Do you have any good suggestions？Thanks

> 同感，但是检查了loss，print了梯度，感觉是没有问题的，是有梯度回传的，但是就是loss降不下去 @libb999 @aoluming 请问解决了吗？