Person re-identification in an open-world task
Research areas
External Member
Description
Abstract: Person re-identification aims to identify an individual in different spatial locations, temporal periods, or under different camera views, considering a large set of candidate targets. The system aims to integrate a human detection and re-identification task using multiple camera modalities (e.g. visible, thermal) under uncontrollable variations in illumination and occlusion. Dataset: adapt and develop dataset (e.g. MARS, RegDB).
Contact: David Ahmedt-Aristizabal