okankop / Efficient-3DCNNs

PyTorch Implementation of "Resource Efficient 3D Convolutional Neural Networks", codes and pretrained models.
MIT License
777 stars 151 forks source link

Jester Pretrained model loading #25

Closed ahkarami closed 4 years ago

ahkarami commented 4 years ago

Dear @okankop, For the pretrained Jester models, I must use the BGR images/frames of range [0, 255] and with the below mean & std normalization (i.e., Kinetics mean)?
mean=[110.63666788, 103.16065604, 96.29023126], std=[1, 1, 1]

okankop commented 4 years ago

Yes you are right

ahkarami commented 4 years ago

Thanks

SatMa34 commented 4 years ago

Thanks

Hi buddy, did you inference with the pretrained model on jester dataset successfully? I use the BGR format input and the mean and std is as you said, but whatever the input is, the top 5 pred label is always: 1,16,24,5,26, the video is always classified to "doing other things", do you know the reason? Thanks