Person Re-Identification on MSMT17

43개 결과 · ⬇ CSV · JSON

mAP

Rank	Model	mAP	Rank-1	Rank-10	Rank-5	Extra Training Data	Paper	Code	Year
1	CLIP-ReID (with re-ranking)	86.7	91.2	–	–		CLIP-ReID: Exploiting Vision-Language Model for Image Re-Identification without Concrete Text Labels	mikel-brostrom/boxmot · syliz517/clip-reid	2022
2	SOLIDER (with re-ranking)	86.5	91.7	–	–	✓	Beyond Appearance: a Semantic Controllable Self-Supervised Learning Framework for Human-Centric Visual Tasks	modelscope/modelscope · tinyvision/SOLIDER · hasanirtiza/Pedestron · +1	2023
3	FlipReID (with re-ranking)	81.3	87.5	–	–		FlipReID: Closing the Gap between Training and Inference in Person Re-Identification	nixingyang/FlipReID	2021
4	ProNet++ (ResNet50+RK)	80	88.2	–	–		Rethinking Person Re-identification from a Projection-on-Prototypes Perspective		2023
5	LDS (ResNet50+RK)	79.09	88.35	–	–		Learning to Disentangle Scenes for Person Re-identification	deropty/LDS	2021
6	SOLIDER (without re-ranking)	77.1	90.7	–	–	✓	Beyond Appearance: a Semantic Controllable Self-Supervised Learning Framework for Human-Centric Visual Tasks	modelscope/modelscope · tinyvision/SOLIDER · hasanirtiza/Pedestron · +1	2023
7	Adaptive L2 Regularization (with re-ranking)	76.7	84.9	–	–		Adaptive L2 Regularization in Person Re-Identification	nixingyang/AdaptiveL2Regularization	2020
8	PCL-CLIP (L_pcl+L_id)	76.1	89.8	96.0	94.7		Prototypical Contrastive Learning-based CLIP Fine-tuning for Object Re-identification	RikoLi/PCL-CLIP	2023
9	CLIP-ReID (without re-ranking)	75.8	89.7	–	–		CLIP-ReID: Exploiting Vision-Language Model for Image Re-Identification without Concrete Text Labels	mikel-brostrom/boxmot · syliz517/clip-reid	2022
10	TransReID-SSL (ViT-B without RK)	75.0	89.5	–	–		Self-Supervised Pre-Training for Transformer-Based Person Re-Identification	DengpanFu/LUPerson · damo-cv/TransReID-SSL · michuanhaohao/transreid-ssl	2021
11	CA-Jaccard	74.1	86.2	–	90.5		CA-Jaccard: Camera-aware Jaccard Distance for Person Re-identification	chen960/ca-jaccard	2023
12	PCL-CLIP (L_pcl)	73.8	89.2	95.8	94.7		Prototypical Contrastive Learning-based CLIP Fine-tuning for Object Re-identification	RikoLi/PCL-CLIP	2023
13	DiP (without RK)	71.8	87.3	–	–		DiP: Learning Discriminative Implicit Parts for Person Re-Identification	siyuch-fdu/DiP	2022
14	TransReID	69.40	86.20	–	–		TransReID: Transformer-based Object Re-Identification	heshuting555/TransReID · damo-cv/transreid · ChristmasStory/TransReID-main · +1	2021
15	Unsupervised Pre-training (ResNet101+MGN)	68.8	86.6	–	–	✓	Unsupervised Pre-training for Person Re-identification	DengpanFu/LUPerson	2020
16	Weakly Supervised Pre-training (ResNet50+MGN)	68.0	86.0	–	–		Large-Scale Pre-training for Person Re-identification with Noisy Labels	DengpanFu/LUPerson · dengpanfu/luperson-nl	2022
16	FlipReID (without re-ranking)	68.0	85.6	–	–		FlipReID: Closing the Gap between Training and Inference in Person Re-Identification	nixingyang/FlipReID	2021
18	CLIP-ReID Baseline + UFFM +AMC	67.6	83.8	–	–		Enhancing person re-identification via Uncertainty Feature Fusion Method and Auto-weighted Measure Combination	chequanghuy/Enhancing-Person-Re-Identification-via-UFFM-and-AMC	2024
19	Deep Miner (w/o ReRank)	67.30	85.60	–	–		Deep Miner: A Deep and Multi-branch Network which Mines Rich and Diverse Features for Person Re-identification	robin-schneider/cicy-fourfolds	2021
19	UniHCP (finetune)	67.3	–	–	–		UniHCP: A Unified Model for Human-Centric Perceptions	opengvlab/unihcp	2023