Adaptive Calibration for Fair and Performant Facial Recognition

2026-06-03

AI Summary

This work addresses the misalignment between cosine similarity and matching probability in face recognition, which exhibits region-dependent discrepancies that lead to calibration bias and fairness issues across demographic groups. To resolve this without relying on demographic metadata, the authors propose an adaptive calibration method that leverages local context in the embedding space to construct region-specific probability mappings. This approach transforms cosine similarities of normalized embeddings into well-calibrated match probabilities. Evaluated across multiple pre-trained models and standard face verification benchmarks, the method consistently improves both calibration accuracy and group fairness on diverse datasets, avoiding the performance degradation commonly associated with traditional “flattening” calibration strategies.

Abstract

We introduce Adaptive Calibration (AC), a novel calibration strategy for facial recognition that maps cosine similarity between normalized embeddings to well-calibrated probabilities. By incorporating local context into calibration, Adaptive Calibration corrects for a fundamental mismatch in cosine similarity, whereby the same distance can correspond to different match probabilities in different embedding regions. Our approach improves both overall performance and results in a fairer calibration without requiring demographic metadata. Our approach consistently dominates existing methods both on accuracy and fairness metrics across a variety of pretrained models and standard benchmarks. AC provides a practical solution for equitable facial recognition, without requiring demographic group annotations, and while improving overall performance. Unlike existing approaches, our method provides continuous, region-specific calibration that avoids "leveling down" where fairness comes at the cost of degraded performance for some groups.

Problem

• facial recognition
• calibration
• fairness
• cosine similarity
• embedding

Innovation

• Adaptive Calibration
• facial recognition
• fairness
• probability calibration
• cosine similarity