Abstract
Numerous face frontalization methods based on 3D Morphable Model (3DMM) and Generative Adversarial Networks (GAN) have made great progress in multi-view face recognition. However, facial feature analysis and identity discrimination often suffer from failure frontalization results because of monotonous single-domain training and unpredictable input profile faces. To overcome the drawback, we present a novel approach named Well-advised Pose Normalization Network (WAPNN), which leverages multiple domains and extracts features considering their frontalization qualities wisely, to achieve a high accuracy on multi-view face recognition. Through multi-domain datasets, we design an end-to-end facial pose normalization network with adaptive weights on different objectives to exploit potentialities of various profile-front relationships. Meanwhile, the proposed method encourages intra-class compactness and inter-class separability between facial features by introducing quality-aware feature fusion. Experimental analyses show that our method effectively recovers frontal faces with good-quality textures and high identity-preserving, and significantly reduces the impact of various poses on face recognition under both constrained and wild environments.
Highlights
INTRODUCTIONFace recognition has made great progress with the development of deep learning
In recent years, face recognition has made great progress with the development of deep learning
Many approaches have been explored to solve the problems met by multi-view face recognition, they can be categorized into feature-based methods and frontalizationbased methods
Summary
Face recognition has made great progress with the development of deep learning. CNN-based methods [27], [28], especially the methods [29]–[31] using Generative Adversarial Networks (GAN) [32] have been made to overcome the drawbacks of geometry-based methods and achieve much better synthesis performance These black-box models heavily rely on the prior information of the training data instead of 3DMM constraints. In order to improve identity preservation of face frontalization and reduce impacts from synthesis artificialities on feature extraction, we propose a novel deep neural framework, termed as Well-advised Pose Normalization Network (WAPNN), to normalize facial poses and extract concatenated features wisely for improving multi-view face recognition. Datasets for the pose normalization network. 3) Quality assessment and feature fusion are deployed to reduce adverse impacts from unsatisfactory syntheses. 4) Comprehensive experiments on different benchmarks demonstrates that our approach can improve pose-invariant face recognition significantly, and achieve superior performance over other state-ofthe-art methods
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
