Uncertainty-Aware Pedestrian Re-Identification in Autonomous Driving Perception Pipelines

Pedestrian re-identification models trained on one city often degrade when deployed in another due to changes in camera placement, lighting, and pedestrian appearance patterns. Building on CLIP-based uncertainty modal modeling, this work proposes an uncertainty-weighted adaptation strategy that transfers vision–language embeddings across cities while down weighting unreliable pseudo-labels. The method combines (i) entropy-based sample filtering, (ii) uncertainty-aware class prototype refinement, and (iii) consistency regularization between image and text embeddings. Experiments are conducted on cross-city splits constructed from autonomous driving data, totaling 320,000 images and 38,000 identities across three urban domains. Compared with OSNet, TransReID, and CLIP-derived ReID baselines under standard unsupervised domain adaptation settings, the proposed method improves target-domain mAP by 4.0%–6.3% and rank-1 accuracy by 3.2%–5.1%, with the largest gains observed in nighttime and rain subsets.