mlsanityData to 13 Oct 2026

Distilling Facial Knowledge With Teacher-Tasks: Semantic-Segmentation-Features For Pose-Invariant Face-Recognition

arXiv 2209.01115 ↗First posted 2 Sep 2022cs.CVcited by 0

On the map: Knowledge distillation → General knowledge distillation methods

Abstract (arXiv)

This paper demonstrates a novel approach to improve face-recognition pose-invariance using semantic-segmentation features. The proposed Seg-Distilled-ID network jointly learns identification and semantic-segmentation tasks, where the segmentation task is then "distilled" (MobileNet encoder). Performance is benchmarked against three state-of-the-art encoders on a publicly available data-set emphasizing head-pose variations. Experimental evaluations show the Seg-Distilled-ID network shows notable robustness benefits, achieving 99.9% test-accuracy in comparison to 81.6% on ResNet-101, 96.1% on VGG-19 and 96.3% on InceptionV3. This is achieved using approximately one-tenth of the top encoder's inference parameters. These results demonstrate distilling semantic-segmentation features can efficiently address face-recognition pose-invariance.

Key facts. No citing papers in the map yet (data through 13 Oct 2026). Placed in General knowledge distillation methods, one of 12 directions in Knowledge distillation. Builds on 13 of its 29 references that are in the map. First posted on arXiv 2 Sep 2022.

Builds on

13 of 29 references.

References · 13

  1. 2015Deep Residual Learning for Image Recognitioncited by 51,950
  2. 2015U-Net: Convolutional Networks for Biomedical Image Segmentationcited by 19,656
  3. 2014Very Deep Convolutional Networks for Large-Scale Image Recognitioncited by 15,321
  4. 2015Rethinking the Inception Architecture for Computer Visioncited by 5,457
  5. 2018MobileNetV2: Inverted Residuals and Linear Bottleneckscited by 4,786
  6. 2016Image-to-Image Translation with Conditional Adversarial Networkscited by 4,516
  7. 2016DeepLab: Semantic Image Segmentation with Deep Convolutional Nets, Atrous Convolution, and Fully Connected CRFscited by 3,240
  8. 2015FaceNet: A Unified Embedding for Face Recognition and Clusteringcited by 2,838
  9. 2016Large-Margin Softmax Loss for Convolutional Neural Networkscited by 296
  10. 2016HyperFace: A Deep Multi-task Learning Framework for Face Detection, Landmark Localization, Pose Estimation, and Gender Recognitioncited by 103
  11. 2017Multi-Task Convolutional Neural Network for Pose-Invariant Face Recognitioncited by 24
  12. 2016Face Alignment In-the-Wild: A Surveycited by 6
1 more
  1. 2018GridFace: Face Rectification via Learning Local Homography Transformationscited by 6

Built on by

Not cited yet.