# Teacher-Student Training and Triplet Loss for Facial Expression Recognition under Occlusion

arXiv: https://arxiv.org/abs/2008.01003 · HTML: https://mlsanity.com/paper/2008.01003/

## Key facts

- Cited by 11 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 3 Aug 2020.

## Built on by (most cited first, 11 of 11)

- [Boosting Facial Expression Recognition by A Semi-Supervised Progressive Teacher](https://mlsanity.com/paper/2205.14361/) (2022)
- [Teacher-Student Training and Triplet Loss to Reduce the Effect of Drastic Face Occlusion](https://mlsanity.com/paper/2111.10561/) (2021)
- [A realistic approach to generate masked faces applied on two novel masked face recognition data sets](https://mlsanity.com/paper/2109.01745/) (2021)
- [Triplet Loss-less Center Loss Sampling Strategies in Facial Expression Recognition Scenarios](https://mlsanity.com/paper/2302.04108/) (2023)
- [Learning Using Generated Privileged Information by Text-to-Image Diffusion Models](https://mlsanity.com/paper/2309.15238/) (2023)
- [EmojiHeroVR: A Study on Facial Expression Recognition under Partial Occlusion from Head-Mounted Displays](https://mlsanity.com/paper/2410.03331/) (2024)
- [Region attention and graph embedding network for occlusion objective class-based micro-expression recognition](https://mlsanity.com/paper/2107.05904/) (2021)
- [Class Anchor Margin Loss for Content-Based Image Retrieval](https://mlsanity.com/paper/2306.00630/) (2023)
- [Unimodal and Multimodal Static Facial Expression Recognition for Virtual Reality Users with EmoHeVRDB](https://mlsanity.com/paper/2412.11306/) (2024)
- [Audio Sentiment Analysis via Distillation and Cross-Modal Integration of Generated Multilingual Transcripts](https://mlsanity.com/paper/2607.06611/) (2026)
- [DF-MoE: Generalizable Deepfake Detection via Multimodal Sparse Mixture-of-Experts](https://mlsanity.com/paper/2608.23363/) (2026)

## Abstract (arXiv)

In this paper, we study the task of facial expression recognition under strong occlusion. We are particularly interested in cases where 50% of the face is occluded, e.g. when the subject wears a Virtual Reality (VR) headset. While previous studies show that pre-training convolutional neural networks (CNNs) on fully-visible (non-occluded) faces improves the accuracy, we propose to employ knowledge distillation to achieve further improvements. First of all, we employ the classic teacher-student training strategy, in which the teacher is a CNN trained on fully-visible faces and the student is a CNN trained on occluded faces. Second of all, we propose a new approach for knowledge distillation based on triplet loss. During training, the goal is to reduce the distance between an anchor embedding, produced by a student CNN that takes occluded faces as input, and a positive embedding (from the same class as the anchor), produced by a teacher CNN trained on fully-visible faces, so that it becomes smaller than the distance between the anchor and a negative embedding (from a different class than the anchor), produced by the student CNN. Third of all, we propose to combine the distilled embeddings obtained through the classic teacher-student strategy and our novel teacher-student strategy based on triplet loss into a single embedding vector. We conduct experiments on two benchmarks, FER+ and AffectNet, with two CNN architectures, VGG-f and VGG-face, showing that knowledge distillation can bring significant improvements over the state-of-the-art methods designed for occluded faces in the VR setting.