Chen-Lin Zhang

Thanks for your reply! I just use the same FFmpeg version. But I find that the format is rawvideo in the ./reencode.sh. and the decoder flag is `AV_CODEC_ID_MPEG4` in coviar_data_loader.c...

Here's the output of ffprobe (with user installed git version): ``` zhangcl2@lamda6-GPU4:/data/zhangcl2/UCF-101-mpeg/ApplyEyeMakeup$ /data/zhangcl2/ffmpeg/bin/ffprobe v_ApplyEyeMakeup_g01_c01.mp4 ffprobe version N-90418-g74c6a6d Copyright (c) 2007-2018 the FFmpeg developers built with gcc 4.8 (Ubuntu 4.8.4-2ubuntu1~14.04.3) configuration:...

To who have the same problem with me, I use these strategies: 1. Change code to `pCodec = avcodec_find_decoder(AV_CODEC_ID_RAWVIDEO);` from ` // pCodec = avcodec_find_decoder(AV_CODEC_ID_H264);` 2. Change reencode.sh `/data/zhangcl2/ffmpeg/bin/ffmpeg -i...

I used PyTorch 0.3.1. and the error is about (W.data, not in _u.data).

For example: with such lines of FFT: ` class testFFT(nn.Module): def __init__(self): super(testFFT, self).__init__() def forward(self,x): new_out = afft.Fft(x, Variable(torch.zeros(x.size())).cuda()) return x x = Variable(torch.Tensor(64*14*14,8192)).cuda() temp_fft = torch.nn.DataParallel(testFFT()) temp_fft.cuda() out=temp_fft(x)...

> > We are still occupied with the ECCV supplementary materials. The plan is to update the repo (for customized training) in the upcoming weeks. > > can you share...

> > > > We are still occupied with the ECCV supplementary materials. The plan is to update the repo (for customized training) in the upcoming weeks. > > >...

For the action recognition task, you can directly use the average of RGB prob and Flow prob as the final action prediction. For the action detection task, the common practice...

Hi, Thank you for your interest in our project! Hopefully we will release that tutorial soon (no later than the ECCV conference).

> > > > > > > We are still occupied with the ECCV supplementary materials. The plan is to update the repo (for customized training) in the upcoming weeks....