AdaSin: Enhancing Hard Sample Metrics with Dual Adaptive Penalty for Face Recognition

2025-03-05

AI Summary

Existing face recognition methods suffer from inaccurate hard-sample measurement and inflexible loss functions that cannot dynamically characterize sample difficulty. Method: This paper proposes AdaSin, a novel loss function that employs the sine of the angle between embedding features and class centers as a differentiable, geometrically interpretable dynamic difficulty metric. AdaSin introduces a dual adaptive penalty mechanism—separately modulating positive and negative cosine similarities for hard samples—and integrates curriculum learning to adaptively refine classification boundaries. Contribution/Results: By jointly enhancing intra-class compactness and inter-class separability, AdaSin achieves state-of-the-art performance across eight mainstream face recognition benchmarks. It significantly improves discrimination on challenging samples, including those with large pose variations, occlusions, and low-quality imaging artifacts.

Technology Category

• Computer Vision: Adversarial Attacks & Robustness
• Machine Learning: Adversarial Learning & Robustness
• Search and Optimization: Learning to Search

Application Category

• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Search and Retrieval-Augmented AI: Web evaluation methodologies and metrics
• Security and Privacy: Large-scale security measurements

Abstract

In recent years, the emergence of deep convolutional neural networks has positioned face recognition as a prominent research focus in computer vision. Traditional loss functions, such as margin-based, hard-sample mining-based, and hybrid approaches, have achieved notable performance improvements, with some leveraging curriculum learning to optimize training. However, these methods often fall short in effectively quantifying the difficulty of hard samples. To address this, we propose Adaptive Sine (AdaSin) loss function, which introduces the sine of the angle between a sample's embedding feature and its ground-truth class center as a novel difficulty metric. This metric enables precise and effective penalization of hard samples. By incorporating curriculum learning, the model dynamically adjusts classification boundaries across different training stages. Unlike previous adaptive-margin loss functions, AdaSin introduce a dual adaptive penalty, applied to both the positive and negative cosine similarities of hard samples. This design imposes stronger constraints, enhancing intra-class compactness and inter-class separability. The combination of the dual adaptive penalty and curriculum learning is guided by a well-designed difficulty metric. It enables the model to focus more effectively on hard samples in later training stages, and lead to the extraction of highly discriminative face features. Extensive experiments across eight benchmarks demonstrate that AdaSin achieves superior accuracy compared to other state-of-the-art methods.

Problem

• Improves hard sample difficulty quantification in face recognition
• Introduces dual adaptive penalty for better intra-class and inter-class separation
• Enhances discriminative feature extraction through curriculum learning

Innovation

• Introduces AdaSin loss with sine-based difficulty metric
• Uses dual adaptive penalty for cosine similarities
• Combines curriculum learning for dynamic boundary adjustment