Abstract: Highlights•We introduce Cerberus, a novel attribute-based person re-ID framework.•Cerberus handles reID, PAR, and (partial) APS tasks, using a unified model.•We encourage discerning differences between similar persons via semantic guidance.•We propose a regularization method for improved generalization on unseen data.•Cerberus compete well on reID, PAR, and APS benchmarks without any fine-tuning.
Loading