Attention-Based Image-to-Video Translation for Synthesizing Facial Expression Using GAN

Figure 11

The generated frames with MAEGAN + SA for the four categories of expression on the local dataset (sample 2).