Occluded Face In-painting Using Generative Adversarial Networks—A Review

• Victor Ivamoto USP https://orcid.org/0000-0001-5463-3645
• Rodolfo Simões USP https://orcid.org/0000-0001-6989-1233
• Bruno Kemmer USP https://orcid.org/0000-0002-6830-3281
• Clodoaldo Lima USP https://orcid.org/0000-0003-3381-5348

Resumo

Face image de-occlusion and inpainting is a challenging problem in computer vision with several practical uses and is employed in many image preprocessing applications. The impressive results achieved by generative adversarial networks in image processing increased the attention of the scientific community in recent years around facial de-occlusion and inpainting. Recent network architecture developments are the two-stage networks using coarse to fine approach, landmarks, semantic segmentation map, and edge maps that guide the inpainting process. Moreover, improved convolutions enlarge the receptive field and filter the values passed to the next layer, and attention layers create relationships between local and distant information. This article presents a brief review of recent developments in GAN-based techniques for de-occlusion and inpainting of face images. In addition, it describes and analyzes network architectures and building blocks. Finally, we identify current limitations and propose directions for future research.