Harnessing the Power of MLLMs for Transferable Text-to-Image Person ReID

arXiv:2405.04940v1 Announce Type: new
Abstract: Text-to-image person re-identification (ReID) retrieves pedestrian images according to textual descriptions. Manually annotating textual descriptions is time-consuming, restricting the scale of existing datasets and therefore the generalization ability of ReID models. As a result, we study the transferable text-to-image ReID problem, where we train a model on our proposed large-scale database and directly deploy it to various datasets for evaluation. We obtain substantial training data via Multi-modal Large Language Models (MLLMs). Moreover, we identify and …

abstract arxiv cs.cv datasets identification image images mllms pedestrian person power scale study text text-to-image textual train type