Person Re Identification On Msmt17

评估指标

Rank-1

mAP

评测结果

各个模型在此基准测试上的表现结果

			Paper Title	Repository
CLIP-ReID (with re-ranking)	91.1	86.7	CLIP-ReID: Exploiting Vision-Language Model for Image Re-Identification without Concrete Text Labels
SOLIDER (with re-ranking)	91.7	86.5	Beyond Appearance: a Semantic Controllable Self-Supervised Learning Framework for Human-Centric Visual Tasks
FlipReID (with re-ranking)	87.5	81.3	FlipReID: Closing the Gap between Training and Inference in Person Re-Identification
ProNet++ (ResNet50+RK)	88.2	80	Rethinking Person Re-identification from a Projection-on-Prototypes Perspective	-
LDS (ResNet50+RK)	88.35	79.09	Learning to Disentangle Scenes for Person Re-identification
SOLIDER (without re-ranking)	90.7	77.1	Beyond Appearance: a Semantic Controllable Self-Supervised Learning Framework for Human-Centric Visual Tasks
Adaptive L2 Regularization (with re-ranking)	84.9	76.7	Adaptive L2 Regularization in Person Re-Identification
PCL-CLIP (L_pcl+L_id)	89.8	76.1	Prototypical Contrastive Learning-based CLIP Fine-tuning for Object Re-identification
CLIP-ReID (without re-ranking)	89.7	75.8	CLIP-ReID: Exploiting Vision-Language Model for Image Re-Identification without Concrete Text Labels
TransReID-SSL (ViT-B without RK)	89.5	75.0	Self-Supervised Pre-Training for Transformer-Based Person Re-Identification
CA-Jaccard	86.2	74.1	CA-Jaccard: Camera-aware Jaccard Distance for Person Re-identification
PCL-CLIP (L_pcl)	89.2	73.8	Prototypical Contrastive Learning-based CLIP Fine-tuning for Object Re-identification
DiP (without RK)	87.3	71.8	DiP: Learning Discriminative Implicit Parts for Person Re-Identification
TransReID	86.20	69.40	TransReID: Transformer-based Object Re-Identification
Unsupervised Pre-training (ResNet101+MGN)	86.6	68.8	Unsupervised Pre-training for Person Re-identification
FlipReID (without re-ranking)	85.6	68.0	FlipReID: Closing the Gap between Training and Inference in Person Re-Identification
Weakly Supervised Pre-training (ResNet50+MGN)	86.0	68.0	Large-Scale Pre-training for Person Re-identification with Noisy Labels
CLIP-ReID Baseline + UFFM +AMC	83.8	67.6	Enhancing person re-identification via Uncertainty Feature Fusion Method and Auto-weighted Measure Combination
UniHCP (finetune)	-	67.3	UniHCP: A Unified Model for Human-Centric Perceptions
Deep Miner (w/o ReRank)	85.60	67.30	Deep Miner: A Deep and Multi-branch Network which Mines Rich and Diverse Features for Person Re-identification

0 of 43 row(s) selected.