MagFace: A Universal Representation for Face Recognition and Quality Assessment

By Qiang Meng, Shichao Zhao, Zhida Huang, Feng Zhou. Computer Vision and Pattern Recognition, 2021

689 citations recorded by the source corpus.

The performance of face recognition system degrades when the variability of the acquired faces increases. Prior work alleviates this issue by either monitoring the face quality in pre-processing or predicting the data uncertainty along with the face feature. This paper proposes MagFace, a category of losses that learn a universal feature embedding whose magnitude can measure the quality of the given face. Under the new loss, it can be proven that the magnitude of the feature embedding monotonically increases if the subject is more likely to be recognized. In addition, Mag-Face introduces an adaptive mechanism to learn a well-structured within-class feature distributions by pulling easy samples to class centers while pushing hard samples away. This prevents models from overfitting on noisy low-quality samples and improves face recognition in the wild. Extensive experiments conducted on face recognition, quality assessments as well as clustering demonstrate its superiority over state-of-the-arts. The code is available at https://github.com/IrvingMeng/MagFace.

Subfields: Computer Vision and Pattern Recognition, Signal Processing

View Paper: https://www.semanticscholar.org/paper/915bd0363fb1b7d6e6a3a85471a324fde870337b

How researchers see this paper

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

In Computer Vision and Pattern Recognition

1. Weiyang Liu
2. Adrian Weller
3. Anil K. Jain
4. Bhiksha Raj
5. Yandong Wen
6. Rita Singh
7. Minchul Kim
8. Xiaoming Liu

In Signal Processing

1. Weiyang Liu
2. Anil K. Jain
3. Minchul Kim
4. Xiaoming Liu
5. Adrian Weller
6. Rita Singh
7. Bhiksha Raj
8. Feng Zhou

Citations in the corpus

• AdaFace: Quality Adaptive Margin for Face Recognition (2022, background): “MagFace [23] explores the idea of applying different margins based on recognizability.”
• SphereFace Revived: Unifying Hyperspherical Face Recognition (2021, background): “[45] combines the sample quality to hyperspherical FR through a customized characteristic function that is dependent on the feature magnitude.”
• Learning Compatible Embeddings (2021, background): “Visual search or retrieval systems [36, 37] are widely used in many real-world applications such as face recognition [42, 39, 33, 26, 27], person re-identification [40, 16, 34], car re-identification [19] and image retrieval [2, 12].”
• ElasticFace: Elastic Margin Loss for Deep Face Recognition (2021, methodology): “The great success of softmax loss with penalty margin motivated several works to propose a novel variant of softmax loss [11], [14], [5], [13], [27], [10], [19], [1].”