On the Burstiness of Faces in Set

2025-06-25

AI Summary

In Set-based Face Recognition (SFR), a prevalent “burstiness” phenomenon—where specific identities appear abnormally frequently within sets—degrades model generalization and distorts similarity estimation. This work is the first to systematically characterize burstiness across multiple SFR benchmarks. We propose a three-stage solution: (1) A bursty-face detection strategy leveraging Quickshift++ clustering, feature self-similarity analysis, and generalized max-pooling; (2) A quality-aware dynamic weighting pooling mechanism that suppresses contributions from bursty faces while amplifying those from rare or low-quality instances during both training and inference; (3) An end-to-end differentiable burstiness suppression framework. Extensive experiments demonstrate significant improvements in recognition accuracy and robustness, achieving consistent performance gains across mainstream SFR datasets—including IJB-C, BUPT-Balanced, and BUPT-Global—without architectural modifications or additional supervision.

Technology Category

• Computer Vision: Biometrics, Face, Gesture & Pose
• Data Mining & Knowledge Management: Anomaly/Outlier Detection
• Search and Optimization: Distributed Search

Application Category

• Search and Retrieval-Augmented AI: Web evaluation methodologies and metrics
• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systems

Abstract

Burstiness, a phenomenon observed in text and image retrieval, refers to that particular elements appear more times in a set than a statistically independent model assumes. We argue that in the context of set-based face recognition (SFR), burstiness exists widely and degrades the performance in two aspects: Firstly, the bursty faces, where faces with particular attributes %exist frequently in a face set, dominate the training instances and dominate the training face sets and lead to poor generalization ability to unconstrained scenarios. Secondly, the bursty faces %dominating the evaluation sets interfere with the similarity comparison in set verification and identification when evaluation. To detect the bursty faces in a set, we propose three strategies based on Quickshift++, feature self-similarity, and generalized max-pooling (GMP). We apply the burst detection results on training and evaluation stages to enhance the sampling ratios or contributions of the infrequent faces. When evaluation, we additionally propose the quality-aware GMP that enables awareness of the face quality and robustness to the low-quality faces for the original GMP. We give illustrations and extensive experiments on the SFR benchmarks to demonstrate that burstiness is widespread and suppressing burstiness considerably improves the recognition performance.

Problem

• Burstiness degrades face recognition performance in sets
• Bursty faces dominate training and reduce generalization ability
• Bursty faces interfere with similarity comparison during evaluation

Innovation

• Quickshift++ for bursty face detection
• Feature self-similarity analysis method
• Quality-aware generalized max-pooling technique