中科院计算所视觉信息处理与学习组
中科院计算所视觉信息处理与学习组


News / Events

VIPL's paper on occluded face recognition is accepted by IEEE TIP

Date of publication:2021-11-26      Number of clicks: 10

  Recently, one paper on occluded face recognition is accepted by the journal IEEE TIP. The full name of IEEE TIP is IEEE Transactions on Image Processing, which is an international journal on computer vision and image processing with an impact factor of 10.856 announced in 2021. The paper information is as follow:

  Mingjie He, Jie Zhang, Shiguang Shan, Xiao Liu, Zhongqin Wu, Xilin Chen, "Locality-aware Channel-wise Dropout for Occluded Face Recognition", IEEE Transactions on Image Processing (TIP), 2021. (Accepted)

  Face recognition remains a challenging task in unconstrained scenarios, especially when faces are partially occluded. To improve the robustness against occlusion, augmenting the training images with artificial occlusions has been proved as a useful approach. However, these artificial occlusions are commonly generated by adding a black rectangle or several object templates including sunglasses, scarfs and phones, which cannot well simulate the realistic occlusions. In this paper, based on the argument that the occlusion essentially damages a group of neurons, we propose a novel and elegant occlusion-simulation method via dropping the activations of a group of neurons in some elaborately selected channel. Specifically, we first employ a spatial regularization to encourage each feature channel to respond to local and different face regions. Then, the locality-aware channel-wise dropout (LCD) is designed to simulate occlusions by dropping out a few feature channels. The proposed LCD can encourage its succeeding layers to minimize the intra-class feature variance caused by occlusions, thus leading to improved robustness against occlusion. In addition, we design an auxiliary spatial attention module by learning a channel-wise attention vector to reweight the feature channels, which improves the contributions of non-occluded regions. Extensive experiments on various benchmarks show that the proposed method outperforms state-of-the-art methods with a remarkable improvement.



  

  The main contributions of this paper are summarized as follows:

  1) We propose a novel method to better simulate realistic occlusions by dropping a group of activations in intermediate features. It significantly improves the robustness to occlusions by encouraging the neural network to emphasize on learning discriminative features from the non-occluded face regions.

  2) An auxiliary spatial attention module is designed to improve the contributions of non-occluded regions by adaptively reweighting the feature channels.

  3) Our method significantly outperforms the state-of-the-art methods on IJB-C, LFW and MegaFace benchmarks, especially on the IJB-C dataset with large-scale real-occluded face images.




中科院计算所视觉信息处理与学习组
  • 单位地址:北京海淀区中关村科学院南路6号
  • 邮编:100190
  • 联系电话:010-62600514
  • Email:yi.cheng@vipl.ict.ac.cn
  • Valse

  • 深度学习大讲堂

版权所有 @ 中科院计算所视觉信息处理与学习组 京ICP备05002829号-1 京公网安备1101080060