davide-coccomini / Combining-EfficientNet-and-Vision-Transformers-for-Video-Deepfake-Detection

Code for Video Deepfake Detection model from "Combining EfficientNet and Vision Transformers for Video Deepfake Detection" presented at ICIAP 2021.
https://dl.acm.org/doi/abs/10.1007/978-3-031-06433-3_19
MIT License
237 stars 60 forks source link

ff++dataset structure issue #54

Closed Mrmao233 closed 4 months ago

Mrmao233 commented 4 months ago

The preprocessing process seems to have some problems with the ff++ dataset. The structure of the ff++ dataset I downloaded from the official script is as follows, but there will be various problems following the steps in readme. Could you explain the preprocessing process for the ff++ dataset? Thank you very much. 1.9G ./ff++/original_sequences/youtube/c23/videos 1.9G ./ff++/original_sequences/youtube/c23 1.9G ./ff++/original_sequences/youtube 4.6G ./ff++/original_sequences 1.9G ./ff++/manipulated_sequences/Face2Face/c23/videos 1.9G ./ff++/manipulated_sequences/Face2Face/c23 1.9G ./ff++/manipulated_sequences/Face2Face 1.9G ./ff++/manipulated_sequences/Deepfakes/c23/videos 1.9G ./ff++/manipulated_sequences/Deepfakes/c23 1.9G ./ff++/manipulated_sequences/Deepfakes 4.0K ./ff++/manipulated_sequences/FaceSwap/c23/videos/boxes 1.6G ./ff++/manipulated_sequences/FaceSwap/c23/videos 1.6G ./ff++/manipulated_sequences/FaceSwap/c23 1.6G ./ff++/manipulated_sequences/FaceSwap 1.8G ./ff++/manipulated_sequences/FaceShifter/c23/videos 1.8G ./ff++/manipulated_sequences/FaceShifter/c23 1.8G ./ff++/manipulated_sequences/FaceShifter 1.5G ./ff++/manipulated_sequences/NeuralTextures/c23/videos 1.5G ./ff++/manipulated_sequences/NeuralTextures/c23 1.5G ./ff++/manipulated_sequences/NeuralTextures