search

nii-yamagishilab / ClassNSeg

Implementation and demonstration of the paper: Multi-task Learning for Detecting and Segmenting Manipulated Facial Images and Videos
BSD 3-Clause "New" or "Revised" License
83 stars 12 forks source link

FaceForensics++ dataset has masks in image format rather than .avi #1

Closed nviable closed 5 years ago

nviable commented 5 years ago

Your code seems to expect masks in avi format vidcap_mask = cv2.VideoCapture(os.path.join(f_vid_mask, f))

However the updated FaceForensics++ dataset download code assumes that masks and models are only available in raw image sequence format.

Masks and models are only available for raw images.

I'll be writing up a workaround for this to just use image streams for the masks which will probably be useful for people trying to replicate or extend your work.

honghuy127 commented 5 years ago

Yes, create_dataset_Face2Face.py is just for the old FaceForensics dataset. You can slightly modify create_dataset_Deepfakes.py or create_dataset_FaceSwap.py for the Face2Face part of the FaceForensics++ dataset.