GridFace: Face Rectification via Learning Local Homography Transformations
arXiv:1808.06210
Abstract
In this paper, we propose a method, called GridFace, to reduce facial geometric variations and improve the recognition performance. Our method rectifies the face by local homography transformations, which are estimated by a face rectification network. To encourage the image generation with canonical views, we apply a regularization based on the natural face distribution. We learn the rectification network and recognition network in an end-to-end manner. Extensive experiments show our method greatly reduces geometric variations, and gains significant improvements in unconstrained face recognition scenarios.
arXiv commentTo appear in ECCV 2018
References in corpus (11)
- Deep Learning Face Representation by Joint Identification-Verification
- MS-Celeb-1M: A Dataset and Benchmark for Large-Scale Face Recognition
- Plug & Play Generative Networks: Conditional Iterative Generation of Images in Latent Space
- Amortised MAP Inference for Image Super-resolution
- Beyond Face Rotation: Global and Local Perception GAN for Photorealistic and Identity Preserving Frontal View Synthesis
- Deep Mean-Shift Priors for Image Restoration
- Towards End-to-End Face Recognition through Alignment Learning
- Quality Aware Network for Set to Set Recognition
- Towards Large-Pose Face Frontalization in the Wild
- WarpNet: Weakly Supervised Matching for Single-view Reconstruction
- Reconstruction-Based Disentanglement for Pose-invariant Face Recognition