pytorch-video-recognition
pytorch-video-recognition copied to clipboard
jfzhang95
PyTorch implemented C3D, R3D, R2Plus1D models for video activity recognition.
Anyone try to train from scratch on Ucf101 on C3D? The accuracy keep 1%. I use other models implemented by myself and the accuracy is also 1%. The learning rate...
it seems that you are splitting the ucf101 randomly, but we are not allowed to do like this, right?
https://github.com/jfzhang95/pytorch-video-recognition/blob/master/dataloaders/dataset.py#L198 ``` def randomflip(self, buffer): """Horizontally flip the given image and ground truth randomly with a probability of 0.5.""" if np.random.random() < 0.5: for i, frame in enumerate(buffer): frame =...
hi @jfzhang95 , thanks for your code. I'm trying to train C3D model from scratch using your code. I haven't change any setting. After several epoches, the training loss remains...
for ucf101 I have modified the preprocess code and generated official splits But this code is producing below error. Need help. pytorch-video-recognition/dataloaders/dataset.py", line 294, in crop time_index = np.random.randint(buffer.shape[0] -...
Hello, as mentioned in the description, sklearn is used to split train/val/test data for each dataset. Has anybody tried to train and evaluate C3D model on the split 01 of...
In R2Plus1D_model.py, line 200： https://github.com/jfzhang95/pytorch-video-recognition/blob/ca37de9f69a961f22a821c157e9ccf47a601904d/network/R2Plus1D_model.py#L200 It's actually a convolution of 3 * 7 * 7 with padding=(1, 3, 3), not 1 * 7 * 7！
According to following lines of code from dataset.py, it seems like random flip is triggered only for testing data. https://github.com/jfzhang95/pytorch-video-recognition/blob/a63b85f351be193159eaad011115973bb99b9745/dataloaders/dataset.py#L81-L83 Do I get it wrong, or this is a bug...