Talking Face Generation by Conditional Recurrent Adversarial Network -- Pytorch Version

Pytorch implementation of the paper Talking Face Generation by Conditional Recurrent Adversarial Network accepted in IJCAI 2019. The Tensorflow implentation version will be published soon.

Pre-requisites

The datasets we are trained on are TCD ,LFW, Voxceleb and Obama set. Please ask the permission from the original provider.

If you need help with the preprocessing code, please send me an email.

$ sh run_train_tcd.sh

We are still cleaning the code. Please wait for more details.

Name		Name	Last commit message	Last commit date
Latest commit History 8 Commits
lip_read_training_list		lip_read_training_list
preprocess_LRW		preprocess_LRW
utils		utils
README.md		README.md
dataloader.py		dataloader.py
logger.py		logger.py
loss.py		loss.py
model_D.py		model_D.py
model_G.py		model_G.py
model_G_seq.py		model_G_seq.py
model_LipNet.py		model_LipNet.py
run_test.sh		run_test.sh
run_train_lrw.sh		run_train_lrw.sh
run_train_tcd.sh		run_train_tcd.sh
run_train_vox_rnn.sh		run_train_vox_rnn.sh
test.py		test.py
train.py		train.py