PI-GAN: Learning Pose Independent representations for multiple profile face synthesis

TL;DR

PI-GAN introduces a cyclic shared encoder-decoder framework that learns pose-independent face representations, enabling realistic synthesis of multiple profile views from a single image, with applications in security and vision.

Contribution

It proposes a novel cyclic shared encoder-decoder architecture that improves pose-invariant face synthesis by disentangling pose information from identity features.

Findings

Effective synthesis of multi-view faces from a single image.

Uses high-resolution CFP dataset for evaluation.

Outperforms traditional GANs in pose-invariant face generation.

Abstract

Generating a pose-invariant representation capable of synthesizing multiple face pose views from a single pose is still a difficult problem. The solution is demanded in various areas like multimedia security, computer vision, robotics, etc. Generative adversarial networks (GANs) have encoder-decoder structures possessing the capability to learn pose-independent representation incorporated with discriminator network for realistic face synthesis. We present PIGAN, a cyclic shared encoder-decoder framework, in an attempt to solve the problem. As compared to traditional GAN, it consists of secondary encoder-decoder framework sharing weights from the primary structure and reconstructs the face with the original pose. The primary framework focuses on creating disentangle representation, and secondary framework aims to restore the original face. We use CFP high-resolution, realistic dataset to check the performance.