MTCIR (CLIP L/14)

Reported on 6 benchmarks across 2 tasks · 1 paper

Note: results are matched by exact model name. Different papers may use the same name for different model variants.

Computer Vision6 results

Image RetrievalonFashion IQ
(Recall@10+Recall@50)/2· 2023-11-13
46.42
best: 71.77 (DQU-CIR)
Pretrain like Your Inference: Masked Tuning Improves Zero-Shot Composed Image Retrieval arXiv:2311.07622
Image RetrievalonCIRCO
mAP@10· 2023-11-13
11.63
best: 43.4 (MMRet-MLLM)
Pretrain like Your Inference: Masked Tuning Improves Zero-Shot Composed Image Retrieval arXiv:2311.07622
Image RetrievalonCIRR
R@5· 2023-11-13
54.58
best: 81.08 (CoVR-BLIP-2)
Pretrain like Your Inference: Masked Tuning Improves Zero-Shot Composed Image Retrieval arXiv:2311.07622
Composed Image Retrieval (CoIR)onFashion IQ
(Recall@10+Recall@50)/2· 2023-11-13
46.42
best: 60.57 (CoVR-BLIP-2)
Pretrain like Your Inference: Masked Tuning Improves Zero-Shot Composed Image Retrieval arXiv:2311.07622
Composed Image Retrieval (CoIR)onCIRCO
mAP@10· 2023-11-13
11.63
best: 43.4 (MMRet-MLLM)
Pretrain like Your Inference: Masked Tuning Improves Zero-Shot Composed Image Retrieval arXiv:2311.07622
Composed Image Retrieval (CoIR)onCIRR
R@5· 2023-11-13
54.58
best: 81.08 (CoVR-BLIP-2)
Pretrain like Your Inference: Masked Tuning Improves Zero-Shot Composed Image Retrieval arXiv:2311.07622