GitHub - Polarisjame/DeepFake: Audio-Video MultiModality Learning Task

Data

Please Download data from Kaggle

For more datails, visit my Notion

Download Pretrained VST(Video Swin Transformer) from Video-Swin-Transformer and place .pth under ./checkpoint

python3 train.py --data_root [Data Direction]

Name		Name	Last commit message	Last commit date
Latest commit History 5 Commits
checkpoints/wav2vec2-base-960h		checkpoints/wav2vec2-base-960h
data		data
src		src
.gitattributes		.gitattributes
.gitignore		.gitignore
README.md		README.md
audio_preprocess.py		audio_preprocess.py
config.py		config.py
gpu_mem_track.py		gpu_mem_track.py
modelsize_estimate.py		modelsize_estimate.py
test.py		test.py
test_model.sh		test_model.sh
train.py		train.py
train_model.sh		train_model.sh