Embedded Files
Face Reenactment
Introduction
Introduction
We propose the Pose Adapted Shape Learning (PASL) for large-pose face reenactment. The PASL framework consists of three modules, namely the Pose-Adapted face Encoder (PAE), the Cycle-consistent Shape Generator (CSG), and the Attention-Embedded Generator (AEG). Different from previous approaches that use a single face encoder for identity preservation, we propose multiple Pose-Adapted face Encodes (PAEs) to better preserve facial identity across large poses. Given a source face and a reference face, the CSG generates a recomposed shape that fuses the source identity and reference action in the shape space and meets the cycle consistency requirement. Taking the shape code and the source as inputs, the AEG learns the attention within the shape code and between the shape code and source style to enhance the generation of the desired target face.
Architecture of Pose Adapted Shape Learning (PASL)
Architecture of Pose Adapted Shape Learning (PASL)
PASL model composed of 6 Pose-Adapted Encoders (PAEs), Cycle-consistent Shape Generator (CSG), Attention-Embedded
Generator (AEG). Red dashed lines indicate losses, black dashed lines indicate recomposed shape parameters.
Demo - Full Pose Face Reenactment and Infancy to Seniorhood Age Transformation
Demo - Full Pose Face Reenactment and Infancy to Seniorhood Age Transformation
Demo - Pose Adapted Shape Learning (PASL) for Large-Pose Face Reenactment
Demo - Pose Adapted Shape Learning (PASL) for Large-Pose Face Reenactment
Face Reenactment Source Code
Face Reenactment Source Code
The source code can be downloaded from here.
Page updated
Google Sites
Report abuse