New method improves person search across unseen surveillance cameras
LENS: Likelihood Evaluation and Neighbor Smoothing for Attribute-Based Person Retrieval on Unseen Cameras
Computer Vision and Pattern Recognition
Summary
Finding a person in surveillance images using descriptions like clothing or accessories gets harder when the camera is different from those used before. The authors found that current ways to compare people based on attributes don’t handle this well, especially when attributes are uncertain or when the camera changes. They created LENS, which scores how likely a person matches the exact description and improves results by smoothing scores using similar images. Their approach significantly improves accuracy when tested on multiple datasets and a difficult benchmark.
What this means in practice
- •For security teams: Use enhanced attribute-based search to more accurately find individuals across different surveillance cameras without needing retraining for each new camera.
- •For video analytics developers: Integrate LENS to improve person attribute matching and ranking in large multi-camera systems despite camera changes or attribute uncertainty.
Authors
Nguyen Dang Duy Tan, Vo Thanh Nguyen, Linh T. T. Tran, Huynh Nguyen Phat
Abstract
Attribute-based person retrieval ranks surveillance images from a binary query of semantic attributes, but mean Average Attribute Degree of Match (mADM) drops on unseen cameras as predicted probabilities shift. Common pipelines rank gallery images by their L1 distance to the query, which penalizes a confident single-attribute mismatch no more than small uncertainties across many attributes, while post-hoc recalibration fitted on source validation data fails to transfer because the per-attribute maps fitted on source data do not match those of new cameras. We propose LENS (Likelihood Evaluation and Neighbor Smoothing), combining fine-tuned DINOv3 and CLIP models with a distilled student. LENS regularizes mutually exclusive attribute groups with group cross-entropy in training, scores gallery images by the log-likelihood of the exact query, and smooths gallery logits over nearest neighbors using DINOv3 appearance features from the same forward pass. On held-out datasets, all three changes raise mADM: the likelihood score improves all six retrieval metrics for the main ensemble and mADM for all 48 development models, and training-free kNN smoothing improves all 46 tested development models. On the hidden test set of the UPAR 2027 benchmark, LENS reaches 68.17 mADM.