Composed Image Retrieval (CoIR) on CIRCO

Metric: mAP@10 (higher is better)

LeaderboardDataset

Loading chart...

Results

Sort:

#	Model↕	mAP@10▼	Extra Data	Paper	Date↕	Code
1	MMRet-MLLM	43.4	No	MegaPairs: Massive Data Synthesis For Universal ...	2024-12-19	Code
2	MMRet-Large (CLIP L/14)	40.2	No	MegaPairs: Massive Data Synthesis For Universal ...	2024-12-19	Code
3	SCOT (WACV 2025)	37.88	No	SCOT: Self-Supervised Contrastive Pretraining Fo...	2025-01-12	-
4	SEIZE (CLIP G/14 & GPT-4o)	37.23	No	-	-	Code
5	MagicLens (CoCa L)	35.4	No	MagicLens: Self-Supervised Image Retrieval with ...	2024-03-28	Code
6	MMRet-Base (CLIP B/16)	35	No	MegaPairs: Massive Data Synthesis For Universal ...	2024-12-19	Code
7	IP-CIR + LDRE (CLIP G/14)	34.26	No	Imagine and Seek: Improving Composed Image Retri...	2024-11-24	-
8	SEIZE (CLIP G/14)	33.77	No	-	-	Code
9	LDRE (CLIP G/14)	32.24	No	-	-	Code
10	MagicLens (CoCa B)	32	No	MagicLens: Self-Supervised Image Retrieval with ...	2024-03-28	Code
11	OSrCIR (CLIP G/14)	31.14	No	Reason-before-Retrieve: One-Stage Reflective Cha...	2024-12-15	Code
12	MagicLens (CLIP L)	30.8	No	MagicLens: Self-Supervised Image Retrieval with ...	2024-03-28	Code
13	CoVR-BLIP-2	29.55	No	CoVR-2: Automatic Data Construction for Composed...	2023-08-28	Code
14	ImageScope (CLIP-ViT-L/14)	29.23	No	ImageScope: Unifying Language-Guided Image Retri...	2025-03-13	Code
15	CIReVL (CLIP G/14)	27.59	No	Vision-by-Language for Training-Free Composition...	2023-10-13	Code
16	IP-CIR + LDRE (CLIP L/14)	27.41	No	Imagine and Seek: Improving Composed Image Retri...	2024-11-24	-
17	SEIZE (CLIP L/14)	25.82	No	-	-	Code
18	OSrCIR (CLIP L/14)	25.33	No	Reason-before-Retrieve: One-Stage Reflective Cha...	2024-12-15	Code
19	LDRE (CLIP L/14)	24.03	No	-	-	Code
20	MagicLens (CLIP B)	23.8	No	MagicLens: Self-Supervised Image Retrieval with ...	2024-03-28	Code
21	RTD + LinCIR (CLIP G/14)	22.29	No	An Efficient Post-hoc Framework for Reducing Tas...	2024-06-13	Code
22	LinCIR (CLIP G/14)	21.01	No	Language-only Efficient Training of Zero-shot Co...	2023-12-04	Code
23	CoLLM (Pretrained - CLIP-L/14)	20.8	No	CoLLM: A Large Language Model for Composed Image...	2025-03-25	Code
24	CoLLM (Pretrained - BLIP-L/16)	20.4	No	CoLLM: A Large Language Model for Composed Image...	2025-03-25	Code
25	OSrCIR (CLIP B/32)	19.17	No	Reason-before-Retrieve: One-Stage Reflective Cha...	2024-12-15	Code
26	CIReVL (CLIP L/14)	19.01	No	Vision-by-Language for Training-Free Composition...	2023-10-13	Code
27	LDRE (CLIP B/32)	18.32	No	-	-	Code
28	RTD + LinCIR (CLIP L/14)	18.11	No	An Efficient Post-hoc Framework for Reducing Tas...	2024-06-13	Code
29	CompoDiff (CLIP G/14)	17.71	No	CompoDiff: Versatile Composed Image Retrieval Wi...	2023-03-21	Code
30	CIReVL (CLIP B/32)	15.42	No	Vision-by-Language for Training-Free Composition...	2023-10-13	Code
31	Context-I2W	14.62	No	Context-I2W: Mapping Images to Context-dependent...	2023-09-28	Code
32	iSEARLE-XL (CLIP L/14)	13.61	No	iSEARLE: Improving Textual Inversion for Zero-Sh...	2024-05-05	Code
33	LinCIR (CLIP L/14)	13.58	No	Language-only Efficient Training of Zero-shot Co...	2023-12-04	Code
34	CompoDiff (CLIP L/14)	13.51	No	CompoDiff: Versatile Composed Image Retrieval Wi...	2023-03-21	Code
35	SEARLE-XL (CLIP L/14)	12.73	No	Zero-Shot Composed Image Retrieval with Textual ...	2023-03-27	Code
36	iSEARLE-XL-OTI (CLIP L/14)	12.67	No	iSEARLE: Improving Textual Inversion for Zero-Sh...	2024-05-05	Code
37	MTCIR (CLIP L/14)	11.63	No	Pretrain like Your Inference: Masked Tuning Impr...	2023-11-13	Code
38	iSEARLE (CLIP B/32)	11.24	No	iSEARLE: Improving Textual Inversion for Zero-Sh...	2024-05-05	Code
39	iSEARLE-OTI (CLIP B/32)	10.94	No	iSEARLE: Improving Textual Inversion for Zero-Sh...	2024-05-05	Code
40	SEARLE (CLIP B/32)	9.94	No	Zero-Shot Composed Image Retrieval with Textual ...	2023-03-27	Code
41	Pic2Word	9.51	No	Pic2Word: Mapping Pictures to Words for Zero-sho...	2023-02-06	Code
42	MTCIR (BLIP B/16)	8.03	No	Pretrain like Your Inference: Masked Tuning Impr...	2023-11-13	Code
43	PALAVRA	5.32	No	"This is my unicorn, Fluffy": Personalizing froz...	2022-04-04	Code

#1MMRet-MLLMSOTA
43.4
mAP@10· 2024-12-19
MegaPairs: Massive Data Synthesis For Universal Multimodal Retrieval Code
#2MMRet-Large (CLIP L/14)
40.2
mAP@10· 2024-12-19
MegaPairs: Massive Data Synthesis For Universal Multimodal Retrieval Code
#3SCOT (WACV 2025)
37.88
mAP@10· 2025-01-12
SCOT: Self-Supervised Contrastive Pretraining For Zero-Shot Compositional Retrieval
#4SEIZE (CLIP G/14 & GPT-4o)
37.23
mAP@10
No paperCode
#5MagicLens (CoCa L)SOTA
35.4
mAP@10· 2024-03-28
MagicLens: Self-Supervised Image Retrieval with Open-Ended Instructions Code
#6MMRet-Base (CLIP B/16)
35
mAP@10· 2024-12-19
MegaPairs: Massive Data Synthesis For Universal Multimodal Retrieval Code
#7IP-CIR + LDRE (CLIP G/14)
34.26
mAP@10· 2024-11-24
Imagine and Seek: Improving Composed Image Retrieval with an Imagined Proxy
#8SEIZE (CLIP G/14)
33.77
mAP@10
No paperCode
#9LDRE (CLIP G/14)
32.24
mAP@10
No paperCode
#10MagicLens (CoCa B)
32
mAP@10· 2024-03-28
MagicLens: Self-Supervised Image Retrieval with Open-Ended Instructions Code
#11OSrCIR (CLIP G/14)
31.14
mAP@10· 2024-12-15
Reason-before-Retrieve: One-Stage Reflective Chain-of-Thoughts for Training-Free Zero-Shot Composed Image Retrieval Code
#12MagicLens (CLIP L)
30.8
mAP@10· 2024-03-28
MagicLens: Self-Supervised Image Retrieval with Open-Ended Instructions Code
#13CoVR-BLIP-2SOTA
29.55
mAP@10· 2023-08-28
CoVR-2: Automatic Data Construction for Composed Video Retrieval Code
#14ImageScope (CLIP-ViT-L/14)
29.23
mAP@10· 2025-03-13
ImageScope: Unifying Language-Guided Image Retrieval via Large Multimodal Model Collective Reasoning Code
#15CIReVL (CLIP G/14)
27.59
mAP@10· 2023-10-13
Vision-by-Language for Training-Free Compositional Image Retrieval Code
#16IP-CIR + LDRE (CLIP L/14)
27.41
mAP@10· 2024-11-24
Imagine and Seek: Improving Composed Image Retrieval with an Imagined Proxy
#17SEIZE (CLIP L/14)
25.82
mAP@10
No paperCode
#18OSrCIR (CLIP L/14)
25.33
mAP@10· 2024-12-15
Reason-before-Retrieve: One-Stage Reflective Chain-of-Thoughts for Training-Free Zero-Shot Composed Image Retrieval Code
#19LDRE (CLIP L/14)
24.03
mAP@10
No paperCode
#20MagicLens (CLIP B)
23.8
mAP@10· 2024-03-28
MagicLens: Self-Supervised Image Retrieval with Open-Ended Instructions Code
#21RTD + LinCIR (CLIP G/14)
22.29
mAP@10· 2024-06-13
An Efficient Post-hoc Framework for Reducing Task Discrepancy of Text Encoders for Composed Image Retrieval Code
#22LinCIR (CLIP G/14)
21.01
mAP@10· 2023-12-04
Language-only Efficient Training of Zero-shot Composed Image Retrieval Code
#23CoLLM (Pretrained - CLIP-L/14)
20.8
mAP@10· 2025-03-25
CoLLM: A Large Language Model for Composed Image Retrieval Code
#24CoLLM (Pretrained - BLIP-L/16)
20.4
mAP@10· 2025-03-25
CoLLM: A Large Language Model for Composed Image Retrieval Code
#25OSrCIR (CLIP B/32)
19.17
mAP@10· 2024-12-15
Reason-before-Retrieve: One-Stage Reflective Chain-of-Thoughts for Training-Free Zero-Shot Composed Image Retrieval Code
#26CIReVL (CLIP L/14)
19.01
mAP@10· 2023-10-13
Vision-by-Language for Training-Free Compositional Image Retrieval Code
#27LDRE (CLIP B/32)
18.32
mAP@10
No paperCode
#28RTD + LinCIR (CLIP L/14)
18.11
mAP@10· 2024-06-13
An Efficient Post-hoc Framework for Reducing Task Discrepancy of Text Encoders for Composed Image Retrieval Code
#29CompoDiff (CLIP G/14)SOTA
17.71
mAP@10· 2023-03-21
CompoDiff: Versatile Composed Image Retrieval With Latent Diffusion Code
#30CIReVL (CLIP B/32)
15.42
mAP@10· 2023-10-13
Vision-by-Language for Training-Free Compositional Image Retrieval Code
#31Context-I2W
14.62
mAP@10· 2023-09-28
Context-I2W: Mapping Images to Context-dependent Words for Accurate Zero-Shot Composed Image Retrieval Code
#32iSEARLE-XL (CLIP L/14)
13.61
mAP@10· 2024-05-05
iSEARLE: Improving Textual Inversion for Zero-Shot Composed Image Retrieval Code
#33LinCIR (CLIP L/14)
13.58
mAP@10· 2023-12-04
Language-only Efficient Training of Zero-shot Composed Image Retrieval Code
#34CompoDiff (CLIP L/14)
13.51
mAP@10· 2023-03-21
CompoDiff: Versatile Composed Image Retrieval With Latent Diffusion Code
#35SEARLE-XL (CLIP L/14)
12.73
mAP@10· 2023-03-27
Zero-Shot Composed Image Retrieval with Textual Inversion Code
#36iSEARLE-XL-OTI (CLIP L/14)
12.67
mAP@10· 2024-05-05
iSEARLE: Improving Textual Inversion for Zero-Shot Composed Image Retrieval Code
#37MTCIR (CLIP L/14)
11.63
mAP@10· 2023-11-13
Pretrain like Your Inference: Masked Tuning Improves Zero-Shot Composed Image Retrieval Code
#38iSEARLE (CLIP B/32)
11.24
mAP@10· 2024-05-05
iSEARLE: Improving Textual Inversion for Zero-Shot Composed Image Retrieval Code
#39iSEARLE-OTI (CLIP B/32)
10.94
mAP@10· 2024-05-05
iSEARLE: Improving Textual Inversion for Zero-Shot Composed Image Retrieval Code
#40SEARLE (CLIP B/32)
9.94
mAP@10· 2023-03-27
Zero-Shot Composed Image Retrieval with Textual Inversion Code
#41Pic2WordSOTA
9.51
mAP@10· 2023-02-06
Pic2Word: Mapping Pictures to Words for Zero-shot Composed Image Retrieval Code
#42MTCIR (BLIP B/16)
8.03
mAP@10· 2023-11-13
Pretrain like Your Inference: Masked Tuning Improves Zero-Shot Composed Image Retrieval Code
#43PALAVRASOTA
5.32
mAP@10· 2022-04-04
"This is my unicorn, Fluffy": Personalizing frozen vision-language representations Code