CLIP-ReID: Exploiting Vision-Language Model for Image Re-Identification without Concrete Text Labels

Benchmark Model Rank Results
person-re-identification-on-dukemtmc-reidCLIP-ReID (without re-ranking)#27mAP: 83.1Rank-1: 90.8
person-re-identification-on-market-1501CLIP-ReID (without re-ranking)#54Rank-1: 95.4mAP: 90.5
person-re-identification-on-msmt17CLIP-ReID (with re-ranking)#1mAP: 86.7Rank-1: 91.2
person-re-identification-on-msmt17CLIP-ReID (without re-ranking)#8mAP: 75.8Rank-1: 89.7
vehicle-re-identification-on-vehicleid-smallCLIP-ReID (without re-ranking)#7Rank-1: 85.5Rank-5: 97.2
vehicle-re-identification-on-veri-776CLIP-ReID (without re-ranking)#6mAP: 84.5Rank-1: 97.3