search

Rudrabha / LipGAN

This repository contains the codes for LipGAN. LipGAN was published as a part of the paper titled "Towards Automatic Face-to-Face Translation".
http://cvit.iiit.ac.in/research/projects/cvit-projects/facetoface-translation
MIT License
585 stars 124 forks source link

What kind of video does the 'batch_inference.py' need? #8

Closed OscarBEST closed 4 years ago

OscarBEST commented 4 years ago

Hi, @Rudrabha @prajwalkr thank you for your amazing work!

I'm trying to train a local video using the batch_inference.py, but I encountered the following error even I tried different video sources(all in mp4 format): image

But when I used another video in mp4 format, the batch_inference.py could read the frames in the video but got the following error: image It seems that what batch_inference.py reads are invalid frames.

Thus, I wonder if there are some requirements on the input video so as to continue the trainning.

prajwalkr commented 4 years ago

Hello!

The batch_inference.py file is used for testing/inference and not for training.

In the first case, check if the video is corrupted or if you have specified the path correctly as OpenCV could not find any frames in the given video.

In the second case, dlib could not detect faces in certain frames. This can happen if certain frames contain profile (side) views of the face, rather than near-frontal views.

OscarBEST commented 4 years ago

Thanks! I solved the problem when I changed the video, I guess it was indeed the corrupt video that caused the problem. But I encountered another bug when I tried a different video: image Could you please give me a little bit hints about this? Here is my input: image Thanks a million!

prajwalkr commented 4 years ago

Firstly, the --audio parameter must be a wav file, not mp4.

Also, I have no idea about the first problem you have stated. Does the generated video play in standard media players like VLC?

prajwalkr commented 4 years ago

Closing due to inactivity. Please re-open if necessary.