Attention-Based Partial Face Recognition
Authors:
- 19 September 2021
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 2978-2982
Abstract
Photos of faces captured in unconstrained environments, such as large crowds, still constitute challenges for current face recognition approaches as often faces are occluded by objects or people in the foreground. However, few studies have addressed the task of recognizing partial faces. In this paper, we propose a novel approach to partial face recognition capable of recognizing faces with different occluded areas. We achieve this by combining attentional pooling of a ResNet’s intermediate feature maps with a separate aggregation module. We further adapt common losses to partial faces in order to ensure that the attention maps are diverse and handle occluded parts. Our thorough analysis demonstrates that we outperform all baselines under multiple benchmark protocols, including naturally and synthetically occluded partial faces. This suggests that our method successfully focuses on the relevant parts of the occluded face.Keywords
All Related Versions
This publication has 21 references indexed in Scilit:
- SphereFace: Deep Hypersphere Embedding for Face Recognition2017
- Grid Loss: Detecting Occluded Faces2016
- Multiscale representation for partial face recognition under near infrared illumination2016
- Joint Face Detection and Alignment Using Multitask Cascaded Convolutional NetworksIEEE Signal Processing Letters, 2016
- Partial face detection for continuous authentication2016
- Robust Point Set Matching for Partial Face RecognitionIEEE Transactions on Image Processing, 2016
- Robust partial face recognition using instance-to-class distance2013
- Partial Face Recognition: Alignment-Free ApproachIEEE Transactions on Pattern Analysis and Machine Intelligence, 2012
- An investigation into the use of partial-faces for face recognition2003
- Partial face recognition using radial basis function networks2002