End-to-End Lipreading

The code will be available soon.

The weights for the pre-trained encoders (needed if you would like to re-train the models) are the following:
Frontal30°45°60°Profile

Best single-view models:
Frontal, 30°, 45°, 60°, Profile

Best two-view model:
Frontal + Profile

Best three-view model:
Frontal + 45° + Profile