Using convolutional autoencoders to learn facial action sequences (e.g. from neutral expression to smiling). Semi-supervised triplet training with MultiPIE and MUG facial datasets. (See poster for more)
petersli / simons Goto Github PK
View Code? Open in Web Editor NEWCode for project "Learning Disentangled Representations of Facial Action Sequences Using Convolutional Encoding-Decoding." Research done at SBU Computer Vision Lab.