DynaMix: Generalizable Person Re-identification via Dynamic Relabeling and Mixed Data Sampling
This addresses the challenge of recognizing individuals in surveillance and security applications, but it is incremental as it builds on existing data and methods.
The paper tackles the problem of generalizable person re-identification across unseen cameras by proposing DynaMix, which combines labeled multi-camera and pseudo-labeled single-camera data, resulting in consistent outperformance of state-of-the-art methods.
Generalizable person re-identification (Re-ID) aims to recognize individuals across unseen cameras and environments. While existing methods rely heavily on limited labeled multi-camera data, we propose DynaMix, a novel method that effectively combines manually labeled multi-camera and large-scale pseudo-labeled single-camera data. Unlike prior works, DynaMix dynamically adapts to the structure and noise of the training data through three core components: (1) a Relabeling Module that refines pseudo-labels of single-camera identities on-the-fly; (2) an Efficient Centroids Module that maintains robust identity representations under a large identity space; and (3) a Data Sampling Module that carefully composes mixed data mini-batches to balance learning complexity and intra-batch diversity. All components are specifically designed to operate efficiently at scale, enabling effective training on millions of images and hundreds of thousands of identities. Extensive experiments demonstrate that DynaMix consistently outperforms state-of-the-art methods in generalizable person Re-ID.