Mind the Gap: Bridging Occlusion in Gait Recognition via Residual Gap Correction

2025-07-15

AI Summary

Existing gait recognition methods suffer substantial performance degradation under occlusion, while mainstream occlusion modeling strategies either rely on hard-to-obtain paired occluded–complete data or compromise accuracy on complete gait sequences. To address this, we propose a residual correction framework that explicitly models occluded gait as an adaptive residual deviation from the complete representation—eliminating the need for paired supervision. Our method introduces a residual learning module coupled with an adaptive fusion mechanism, which implicitly compensates for occluded regions while fully preserving global discriminative gait features. This enables unified modeling of both occluded and complete inputs. Extensive experiments on Gait3D, GREW, and BRIAR demonstrate significant improvements in occlusion-robust recognition accuracy, while simultaneously achieving state-of-the-art performance on complete gait sequences.

Technology Category

• Computer Vision: Generative Adversarial Networks (GANs) for Vision
• Intelligent Robots: Multimodal Perception & Sensor Fusion
• Planning, Routing, and Scheduling: Activity and Plan Recognition

Application Category

• User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systems
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs

Abstract

Gait is becoming popular as a method of person re-identification because of its ability to identify people at a distance. However, most current works in gait recognition do not address the practical problem of occlusions. Among those which do, some require paired tuples of occluded and holistic sequences, which are impractical to collect in the real world. Further, these approaches work on occlusions but fail to retain performance on holistic inputs. To address these challenges, we propose RG-Gait, a method for residual correction for occluded gait recognition with holistic retention. We model the problem as a residual learning task, conceptualizing the occluded gait signature as a residual deviation from the holistic gait representation. Our proposed network adaptively integrates the learned residual, significantly improving performance on occluded gait sequences without compromising the holistic recognition accuracy. We evaluate our approach on the challenging Gait3D, GREW and BRIAR datasets and show that learning the residual can be an effective technique to tackle occluded gait recognition with holistic retention.

Problem

• Addressing occlusion challenges in gait recognition
• Improving occluded gait recognition without losing holistic accuracy
• Proposing residual learning for gait signature correction

Innovation

• Residual learning for occluded gait correction
• Adaptive integration of residual gait signatures
• Retains holistic recognition accuracy