Papers › Composed Image Retrieval for Training-Free Domain Conversion

Composed Image Retrieval for Training-Free Domain Conversion

4 Dec 2024arXiv:2412.03297archive 2025-07-28

Nikos Efthymiadis, Bill Psomas, Zakaria Laskar, Konstantinos Karantzalos, Yannis Avrithis, Ondřej Chum, Giorgos Tolias

This work addresses composed image retrieval in the context of domain conversion, where the content of a query image is retrieved in the domain specified by the query text. We show that a strong vision-language model provides sufficient descriptive power without additional training. The query image is mapped to the text input space using textual inversion. Unlike common practice that invert in the continuous space of text tokens, we use the discrete word space via a nearest-neighbor search in a text vocabulary. With this inversion, the image is softly mapped across the vocabulary and is made more robust using retrieval-based augmentation. Database images are retrieved by a weighted ensemble of text queries combining mapped words with the domain text. Our method outperforms prior art by a large margin on standard and newly introduced benchmarks. Code: https://github.com/NikosEfth/freedom

PaperPDFCode

Syntology This paper's arXiv id is not in Syntology's Atlas id list (549,449 ids, as this build read them; the read time is in the build record), so this page links to no Syntology page for it. The list is not the whole graph; Syntology's feed shows papers as they enter the graph.

Code

nikosefth/freedom officialmentioned in papermentioned on GitHubpytorchMITSyntology: no sample linked to this paper (harvested for another paper). Syntology's graph links these samples to Fine-tuning CNN Image Retrieval with No Human Annotation (arXiv:1711.02512; that paper's own run record: 17 ran (of which 0 constructed an object rather than computing a result; 16 with no instrument failure: 1 honoured, 0 violated, 15 with no contract checked; 1 where Syntology's instrument failed) · 7 unverified); nothing checks that they implement this paper's method. report

Repository list and official/mentioned flags are the archive's, frozen 2025-07-28. Reachability, where shown, is from one Syntology probe window (2026-09-16 to 2026-09-18); repositories not probed show nothing. GitHub stars are not tracked.

A paper named beside a repository with no sample linked to this paper is shown with that paper's own run record, not this paper's: “ran” means executed on a synthesized input, not that the code is correct, and “where Syntology's instrument failed” counts failures of Syntology's instrument, not of the code. Papers are listed in arXiv-id order, at most three per repository.

Code Syntology ran Syntology

Not run by Syntology. Nothing on this page verifies that the listed code works.

Tasks

Image RetrievalLanguage ModelingLanguage ModellingRetrievalZero-Shot Composed Image Retrieval (ZS-CIR)

Results from the paper archive 2025-07-28

TaskDatasetModelMetricValueRank at snapshotLeaderboardReport
Zero-Shot Composed Image Retrieval (ZS-CIR) ImageNet-R FreeDom (CLIP-L/14) mAP 29.91 #15 of 20 Archive leaderboard report
Zero-Shot Composed Image Retrieval (ZS-CIR) ImageNet-R SEARLE (CLIP-L/14) mAP 14.04 #16 of 20 Archive leaderboard report
Zero-Shot Composed Image Retrieval (ZS-CIR) ImageNet-R CompoDiff (CLIP-L/14) mAP 12.88 #17 of 20 Archive leaderboard report
Zero-Shot Composed Image Retrieval (ZS-CIR) ImageNet-R WeiCom (CLIP-L/14) mAP 10.47 #18 of 20 Archive leaderboard report
Zero-Shot Composed Image Retrieval (ZS-CIR) ImageNet-R MagicLens (CLIP-L/14) mAP 9.13 #19 of 20 Archive leaderboard report
Zero-Shot Composed Image Retrieval (ZS-CIR) ImageNet-R Pic2Word (CLIP-L/14) mAP 7.88 #20 of 20 Archive leaderboard report
Zero-Shot Composed Image Retrieval (ZS-CIR) Large Time Lags Location (LTLL) FreeDom (CLIP-L/14) mAP 33.24 #1 of 6 Archive leaderboard report
Zero-Shot Composed Image Retrieval (ZS-CIR) Large Time Lags Location (LTLL) WeiCom (CLIP-L/14) mAP 26.6 #2 of 6 Archive leaderboard report
Zero-Shot Composed Image Retrieval (ZS-CIR) Large Time Lags Location (LTLL) SEARLE (CLIP-L/14) mAP 25.46 #3 of 6 Archive leaderboard report
Zero-Shot Composed Image Retrieval (ZS-CIR) Large Time Lags Location (LTLL) MagicLens (CLIP-L/14) mAP 24.21 #4 of 6 Archive leaderboard report
Zero-Shot Composed Image Retrieval (ZS-CIR) Large Time Lags Location (LTLL) CompoDiff (CLIP-L/14) mAP 21.61 #5 of 6 Archive leaderboard report
Zero-Shot Composed Image Retrieval (ZS-CIR) Large Time Lags Location (LTLL) Pic2Word (CLIP-L/14) mAP 21.27 #6 of 6 Archive leaderboard report
Zero-Shot Composed Image Retrieval (ZS-CIR) MiniDomainNet FreeDom (CLIP-L/14) mAP 37.27 #1 of 6 Archive leaderboard report
Zero-Shot Composed Image Retrieval (ZS-CIR) MiniDomainNet CompoDiff (CLIP-L/14) mAP 22.95 #2 of 6 Archive leaderboard report
Zero-Shot Composed Image Retrieval (ZS-CIR) MiniDomainNet SEARLE (CLIP-L/14) mAP 21.78 #3 of 6 Archive leaderboard report
Zero-Shot Composed Image Retrieval (ZS-CIR) MiniDomainNet MagicLens (CLIP-L/14) mAP 20.06 #4 of 6 Archive leaderboard report
Zero-Shot Composed Image Retrieval (ZS-CIR) MiniDomainNet Pic2Word (CLIP-L/14) mAP 12 #5 of 6 Archive leaderboard report
Zero-Shot Composed Image Retrieval (ZS-CIR) MiniDomainNet WeiCom (CLIP-L/14) mAP 8.52 #6 of 6 Archive leaderboard report
Zero-Shot Composed Image Retrieval (ZS-CIR) NICO++ FreeDom (CLIP-L/14) mAP 26.1 #1 of 6 Archive leaderboard report
Zero-Shot Composed Image Retrieval (ZS-CIR) NICO++ MagicLens (CLIP-L/14) mAP 19.66 #2 of 6 Archive leaderboard report
Zero-Shot Composed Image Retrieval (ZS-CIR) NICO++ SEARLE (CLIP-L/14) mAP 15.13 #3 of 6 Archive leaderboard report
Zero-Shot Composed Image Retrieval (ZS-CIR) NICO++ WeiCom (CLIP-L/14) mAP 10.54 #4 of 6 Archive leaderboard report
Zero-Shot Composed Image Retrieval (ZS-CIR) NICO++ CompoDiff (CLIP-L/14) mAP 10.32 #5 of 6 Archive leaderboard report
Zero-Shot Composed Image Retrieval (ZS-CIR) NICO++ Pic2Word (CLIP-L/14) mAP 9.76 #6 of 6 Archive leaderboard report

Ranks are positions in the archive's leaderboards as they stood at the 2025-07-28 snapshot. Results published since then are not among these rows, so a rank here is not a current standing.

Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections