Composed Image Retrieval (CoIR) on CIRR

Metric: R@10 (higher is better)

LeaderboardDataset

Loading chart...

Results

Sort:

#	Model↕	R@10▼	Extra Data	Paper	Date↕	Code
1	CoLLM (finetuned - BLIP-L/16)	84.7	No	CoLLM: A Large Language Model for Composed Image...	2025-03-25	Code
2	CoVR-BLIP-2	83.95	No	CoVR-2: Automatic Data Construction for Composed...	2023-08-28	Code
3	IP-CIR + LDRE (CLIP G/14)	80	No	Imagine and Seek: Improving Composed Image Retri...	2024-11-24	-
4	SEIZE (CLIP G/14)	79.42	No	-	-	Code
5	CoLLM (Pretrained - BLIP-L/16)	78.6	No	CoLLM: A Large Language Model for Composed Image...	2025-03-25	Code
6	ImageScope (CLIP-ViT-L/14)	78.05	No	ImageScope: Unifying Language-Guided Image Retri...	2025-03-13	Code
7	MagicLens (CoCa L)	77.9	No	MagicLens: Self-Supervised Image Retrieval with ...	2024-03-28	Code
8	OSrCIR (CLIP G/14)	77.33	No	Reason-before-Retrieve: One-Stage Reflective Cha...	2024-12-15	Code
9	SCOT (WACV 2025)	74.48	No	SCOT: Self-Supervised Contrastive Pretraining Fo...	2025-01-12	-
10	MagicLens (CLIP L)	74.4	No	MagicLens: Self-Supervised Image Retrieval with ...	2024-03-28	Code
11	CoLLM (Pretrained - CLIP-L/14)	72.8	No	CoLLM: A Large Language Model for Composed Image...	2025-03-25	Code
12	IP-CIR + LDRE (CLIP L/14)	71.21	No	Imagine and Seek: Improving Composed Image Retri...	2024-11-24	-
13	OSrCIR (CLIP L/14)	69.86	No	Reason-before-Retrieve: One-Stage Reflective Cha...	2024-12-15	Code
14	OSrCIR (CLIP B/32)	68.19	No	Reason-before-Retrieve: One-Stage Reflective Cha...	2024-12-15	Code

#1CoLLM (finetuned - BLIP-L/16)SOTA
84.7
R@10· 2025-03-25
CoLLM: A Large Language Model for Composed Image Retrieval Code
#2CoVR-BLIP-2SOTA
83.95
R@10· 2023-08-28
CoVR-2: Automatic Data Construction for Composed Video Retrieval Code
#3IP-CIR + LDRE (CLIP G/14)
80
R@10· 2024-11-24
Imagine and Seek: Improving Composed Image Retrieval with an Imagined Proxy
#4SEIZE (CLIP G/14)
79.42
R@10
No paperCode
#5CoLLM (Pretrained - BLIP-L/16)
78.6
R@10· 2025-03-25
CoLLM: A Large Language Model for Composed Image Retrieval Code
#6ImageScope (CLIP-ViT-L/14)
78.05
R@10· 2025-03-13
ImageScope: Unifying Language-Guided Image Retrieval via Large Multimodal Model Collective Reasoning Code
#7MagicLens (CoCa L)
77.9
R@10· 2024-03-28
MagicLens: Self-Supervised Image Retrieval with Open-Ended Instructions Code
#8OSrCIR (CLIP G/14)
77.33
R@10· 2024-12-15
Reason-before-Retrieve: One-Stage Reflective Chain-of-Thoughts for Training-Free Zero-Shot Composed Image Retrieval Code
#9SCOT (WACV 2025)
74.48
R@10· 2025-01-12
SCOT: Self-Supervised Contrastive Pretraining For Zero-Shot Compositional Retrieval
#10MagicLens (CLIP L)
74.4
R@10· 2024-03-28
MagicLens: Self-Supervised Image Retrieval with Open-Ended Instructions Code
#11CoLLM (Pretrained - CLIP-L/14)
72.8
R@10· 2025-03-25
CoLLM: A Large Language Model for Composed Image Retrieval Code
#12IP-CIR + LDRE (CLIP L/14)
71.21
R@10· 2024-11-24
Imagine and Seek: Improving Composed Image Retrieval with an Imagined Proxy
#13OSrCIR (CLIP L/14)
69.86
R@10· 2024-12-15
Reason-before-Retrieve: One-Stage Reflective Chain-of-Thoughts for Training-Free Zero-Shot Composed Image Retrieval Code
#14OSrCIR (CLIP B/32)
68.19
R@10· 2024-12-15
Reason-before-Retrieve: One-Stage Reflective Chain-of-Thoughts for Training-Free Zero-Shot Composed Image Retrieval Code