RobustFace: Adaptive Mining of Noise and Hard Samples for Robust Face Recognitions
Abstract: While margin-based deep face recognition models, such as ArcFace and AdaFace, have achieved remarkable successes over the recent years, they may suffer from degraded performances when encountering training sets corrupted with noises. This is often inevitable when massively large scale datasets need to be dealt with, yet it remains difficult to construct clean enough face datasets under these circumstances. In this paper, we propose a robust deep face recognition model, RobustFace, by combining the advantages of margin-based learning models with the strength of mining-based approaches to effectively mitigate the impact of noises during trainings. Specifically, we introduce a noise-adaptive mining strategy to dynamically adjust the emphasis balance between hard and noise samples by monitoring the model's recognition performances at the batch level to provide optimization-oriented feedback, enabling direct training on noisy datasets without the requirement of pre-training. Extensive experiments validate that our proposed RobustFace achieves competitive performances in comparison with the existing SoTA models when trained with clean datasets. When trained with both real-world and synthetic noisy datasets, RobustFace significantly outperforms the existing models, especially when the synthetic noisy datasets are corrupted with both close-set and open-set noises. While the existing baseline models suffer from an average performance drop of around 40\%, under these circumstances, our proposed still delivers accuracy rates of more than 90\%.
Primary Subject Area: [Content] Vision and Language
Secondary Subject Area: [Content] Vision and Language
Relevance To Conference: This work contributes to multimedia/multimodal processing by addressing the challenge of robust face recognition in the presence of noisy datasets. Face recognition is a crucial component in various multimedia and multimodal processing applications, such as video surveillance, biometric authentication, and human-computer interaction.
The robustness of we proposed RobustFace model is crucial in multimedia/multimodal processing scenarios where datasets are often prone to noise, such as low-quality images, variations in lighting conditions, occlusions, or pose variations. By achieving competitive performances on clean datasets and significantly outperforming existing models on real-world and synthetic noisy datasets, RobustFace provides improved accuracy and reliability in face recognition tasks.
The advancements in robust face recognition offered by we proposed RobustFace contribute to enhancing the performance and reliability of multimedia/multimodal processing systems that rely on accurate and robust face recognition as a fundamental component.
Submission Number: 317
Loading