SOTAVerified

Text-based Person Retrieval with Noisy Correspondence

This is a benchmark about text-based person retrieval with noisy correspondence. All recorded values ​​are the results under 20% noise rate.

Papers

Recently AddedMost HypedMost ActiveNeeds VerificationMost Verified

Showing 1–6 of 6 papers

TitleStatusHype
Noisy-Correspondence Learning for Text-to-Image Person Re-identificationCode1
Cross-Modal Implicit Relation Reasoning and Aligning for Text-to-Image Person RetrievalCode2
Deep Evidential Learning with Noisy Correspondence for Cross-Modal RetrievalCode1
See Finer, See More: Implicit Modality Alignment for Text-based Person RetrievalCode1
Semantically Self-Aligned Network for Text-to-Image Part-aware Person Re-identificationCode1
Learning Transferable Visual Models From Natural Language SupervisionCode2
Show:102550
All datasetsCUHK-PEDESICFG-PEDESRSTPReid

Benchmark Results

#ModelMetricClaimedVerifiedStatus
1RDERank-174.46—Unverified
2DECLRank-170.29—Unverified
3IRRARank-169.74—Unverified
4CLIP-CRank-166.41—Unverified
5IVTRank-158.59—Unverified
6SSANRank-146.52—Unverified
#ModelMetricClaimedVerifiedStatus
1RDERank 166.54—Unverified
2DECLRank 161.95—Unverified
3IRRARank 160.76—Unverified
4CLIP-CRank 155.25—Unverified
5IVTRank 150.21—Unverified
6SSANRank 140.57—Unverified
#ModelMetricClaimedVerifiedStatus
1RDERank 164.45—Unverified
2DECLRank 161.75—Unverified
3IRRARank 158.75—Unverified
4CLIP-CRank 154.45—Unverified
5IVTRank 143.65—Unverified
6SSANRank 135.1—Unverified