Learning Meta Face Recognition in Unseen Domains

By Jianzhu Guo, Xiangyu Zhu, Chenxu Zhao, Dong Cao, Zhen Lei, S. Li. 2020

160 citations recorded by the source corpus.

Face recognition systems are usually faced with unseen domains in real-world applications and show unsatisfactory performance due to their poor generalization. For example, a well-trained model on webface data cannot deal with the ID vs. Spot task in surveillance scenario. In this paper, we aim to learn a generalized model that can directly handle new unseen domains without any model updating. To this end, we propose a novel face recognition method via meta-learning named Meta Face Recognition (MFR). MFR synthesizes the source/target domain shift with a meta-optimization objective, which requires the model to learn effective representations not only on synthesized source domains but also on synthesized target domains. Specifically, we build domain-shift batches through a domain-level sampling strategy and get back-propagated gradients/meta-gradients on synthesized source/target domains by optimizing multi-domain distributions. The gradients and meta-gradients are further combined to update the model to improve generalization. Besides, we propose two benchmarks for generalized face recognition evaluation. Experiments on our benchmarks validate the generalization of our method compared to several baselines and other state-of-the-arts. The proposed benchmarks and code will be available at https://github.com/cleardusk/MFR.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

1. Shengcai Liao
2. Manmohan Chandraker
3. Yi-Hsuan Tsai
4. Mei Wang
5. Zhen Lei
6. Weihong Deng
7. Xiangyu Zhu
8. Tao Mei

In Computer Vision and Pattern Recognition

1. Luc Van Gool
2. Yi Yang
3. Xiaogang Wang
4. Dacheng Tao
5. Weihong Deng
6. Qi Tian
7. Bernt Schiele
8. S. Li

Citations in the corpus

• Meta-Learning with Less Forgetting on Large-Scale Non-Stationary Task Distributions (2022, neutral): “Meta-learning [51] focuses on learning lots of related tasks to acquire common knowledge adaptable to new unseen tasks, and has achieved great success in many computer vision problems [26,61,27].”
• Exclusivity-Consistency Regularized Knowledge Distillation for Face Recognition (2020, background): “Convolutional neural networks (CNNs) have gained impressive success in the recent advanced face recognition systems [47, 24, 12, 13, 53, 45, 48, 44].”
• Deep Face Recognition: A Survey (2018, background): “unbalancedtraining [283], attribute removal [284], [285], [286] and domain adaptation [173], [287], [147].”
• Face Anti-Spoofing with Human Material Perception (2020, background): “In recent years, face recognition [1,2,3] has been widely used in various interactive and payment scene due to its high accuracy and convenience.”
• Towards Fast, Accurate and Stable 3D Dense Face Alignment (2020, background): “, recognition [45,7,26,24,13,51], animation [10], avatar retargeting [9], tracking [50], attribute classification [4,22,21], image restoration [52,12,11], anti-spoofing [49,55,40,54,25].”
• Cross-Domain Similarity Learning for Face Recognition in Unseen Domains (2021, methodology): “In the end, the model parameters are updated by accumulated gradients of both Ls and Lt, as this has been shown to be more effective than the original MAML [2, 18].”
• MagFace: A Universal Representation for Face Recognition and Quality Assessment (2021, background): “To achieve high end-to-end application performances in the second stage, various metric-learning [27, 30] or classification losses [48, 25, 20, 13, 40, 9, 5] emerged in the past few years.”