Pith. sign in

REVIEW 3 major objections 5 minor 65 references

DetailFusion: A Dual-branch Framework with Detail Enhancement for Composed Image Retrieval

T0 review · 3 major / 5 minor · reviewed 2026-08-07 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read DetailFusion claims that pre-training one branch on image-editing triplets and adaptively fusing it with a global branch gives state-of-the-art composed image retrieval on CIRR and FashionIQ.

desk verdict A solid, well-ablated CIR method whose SOTA numbers are compromised by a validation-set checkpoint selection that baselines don't get. read the letter →

arxiv 2505.17796 v1 pith:PZ6CHAJN submitted 2025-05-23 cs.CV cs.AIcs.IR

Yuxin Yang , Yinan Zhou , Yuxin Chen , Ziqi Zhang , Zongyang Ma , Chunfeng Yuan , Bing Li , Lin Song
show 3 more authors
Jun Gao Peng Li Weiming Hu
This is my paper
classification cs.CVcs.AIcs.IR
keywords composedimageretrievaldetailenhancementdual-branchframeworkeditingpre-traininghardnegativeadaptivefeaturecompositorCIRRFashionIQ
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

Composed image retrieval lets a user find an image by pointing at a reference photo and describing a change, like making the cup brown instead of red. The paper argues that previous supervised methods fail on such queries because they fuse image and text at a coarse, global level and never learn to perceive fine-grained visual details or execute small textual modifications. It proposes DetailFusion, a two-branch framework: one branch is dedicated to global semantics, and one detail-oriented inference branch is pre-trained on the InstructPix2Pix image-editing dataset, where each query's reference image is deliberately used as a hard negative during training. A lightweight Adaptive Feature Compositor then learns to combine the two branches' features per query. The paper reports state-of-the-art Recall@K on CIRR and FashionIQ and claims the added detail capability transfers to unseen domains in zero-shot settings.

What carries the argument

The load-bearing object is the Detail-oriented Inference (DI) branch, trained with a detail-oriented contrastive loss in which the reference image itself is added as a hard negative to the denominator, forcing the model to ignore near-duplicate visual similarity and attend to the specific alteration described by the text. The second mechanism is the Adaptive Feature Compositor, which uses cross-attention between the [CLS] tokens of the two branches and their detailed token sets, then forms a convex combination of the global and detail features with a learned ratio and a bridging feature. The design rests on the premise that global and detail branches are complementary, as shown by branch-level ablations where the GM branch excels at global Recall@1 while the DI branch excels at subset retrieval.

What would settle it

Replace the IPr2Pr pre-training with an equal-size sample of the target CIR dataset's own triplets, keeping the same hard-negative loss and the same branch architecture; if Recallsubset@1 on CIRR does not drop, then the paper's attribution of the gain to image-editing priors is falsified.

Watch

Extended reading notes

Core claim

On the paper's own terms, the central claim is that the failure of current supervised CIR models to handle subtle modifications stems from missing targeted training for fine-grained details, and that this can be fixed by a three-stage recipe. First, pre-train a Detail-oriented Inference branch on the InstructPix2Pix image-editing dataset, whose triplets have near-identical reference and target images and short atomic edit texts, using a contrastive loss that treats the reference image as a hard negative. Second, fine-tune that branch together with a Global Feature Matching branch on CIR data, keeping the detail-oriented loss for the detail branch. Third, train a lightweight Adaptive Feature Compositor that fuses global and detail features through cross-attention and a convex combination. The paper claims this yields state-of-the-art results on CIRR and FashionIQ, with gains particularly on the CIRR subset metric that measures fine-grained discrimination, and that the detail capability transfers zero-shot to CIRCO and GeneCIS.

Load-bearing premise

The whole approach rests on the assumption that the kinds of small edits found in an image-editing dataset teach a model the same fine-grained distinctions that composed image retrieval queries ask for; if those editing cues do not transfer, the detail branch's extra training will not help retrieval.

Editorial extensions

If this is right

  • Supervised CIR models can be improved without new retrieval-specific annotations by borrowing an image-editing dataset, since editing triplets encode atomic text-image transformations.
  • Explicitly keeping a detail branch and a global branch separate, then fusing them adaptively, outperforms single-branch fusion; the DI branch alone wins on CIRR subset metrics while the GM branch wins on global Recall@1.
  • The detail capability transfers to unseen domains: zero-shot CIRCO and GeneCIS results improve over the same-backbone baseline, with the largest gains on 'change' subsets.
  • The extra training cost is modest (10-20% over single-stage methods) and inference is about 1.5x a single Q-Former and only 12% slower than SPRC, making the approach practical.
  • Directly mixing the editing dataset with CIR data in a single stage hurts performance; the three-stage coarse-to-fine schedule is necessary for the transfer to work.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • If the atomic detail priors transfer, one could pre-train the detail branch on any large paired image-edit dataset and expect similar gains on CIR benchmarks, suggesting a data-centric alternative to architecture search.
  • The adaptive compositor's convex combination with a learned bridging feature could generalize beyond CIR to other tasks where a query has two granularities, such as visual question answering or image-caption retrieval.
  • The hard-negative reference trick is a general recipe for contrastive retrieval: injecting a near-duplicate query's own reference as a negative forces finer discrimination, and the paper's ablations imply this is what preserves the DI branch's subset performance.
  • A testable extension is to evaluate on a benchmark with edit-type labels to see whether the DI branch learns disentangled atomic edit types, since the current aggregate subset metric does not reveal whether individual transformations are decoupled.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 5 minor

Summary. The paper proposes DetailFusion, a dual-branch framework for supervised composed image retrieval (CIR). A Detail-oriented Inference (DI) branch is pre-trained on the InstructPix2Pix (IPr2Pr) image-editing dataset with a contrastive loss that uses the reference image as a hard negative, then jointly fine-tuned with a Global Feature Matching (GM) branch on CIR datasets. An Adaptive Feature Compositor, trained in a third stage, dynamically fuses the global and detail features. The authors report state-of-the-art results on CIRR and FashionIQ, provide ablations supporting each design choice, and include zero-shot evaluations on CIRCO and GeneCIS, plus computational-cost comparisons. The paper claims that explicitly training one branch on fine-grained visual and textual variations improves supervised CIR at 10-20% additional training cost.

Significance. If the claimed results hold, the work is significant for supervised CIR: it would demonstrate that image-editing triplets, though distributionally different from retrieval data, can be used to train a detail-specialized branch that transfers to CIR benchmarks, and that a lightweight adaptive compositor can combine global and detail features effectively. The paper has several genuine strengths: the ablation of the SPRC-with-same-pretraining control in Table 2(a) directly targets the 'more data' confound, the three-stage training strategy is systematically ablated in Tables 2(b)-2(d), the loss-function variants are compared in Table 2(c), and the cross-domain zero-shot results in Tables 5-6 provide external evidence for the transferability of the detail prior. The main weakness is an experimental-protocol issue: the Stage-1 pre-training checkpoint is selected on the target benchmarks, which is not controlled for in the comparison against baselines, and all numbers are single-run with no error bars. These issues bear directly on the central 'state-of-the-art' claim.

major comments (3)
  1. [Section 4.1, Implementation Details] The text states: 'the best-performing iteration on CIRR and FashionIQ is selected as the initialization for the next stage.' This is selection on the target benchmarks: for FashionIQ, Table 1 reports results on exactly this validation set, so checkpoint selection and evaluation share the same labels; for CIRR, the validation set is also used for ablation and hyperparameter choices. No comparable selection step is described for SPRC or the other baselines. This makes the reported SOTA comparison uncontrolled, and the gains attributed to detail enhancement could partly be a validation-selection artifact. Please report the performance range across Stage-1 iterations, use a selection criterion that does not touch the evaluation set (e.g., an IPr2Pr held-out split or the final checkpoint), and rerun the comparison under that protocol.
  2. [Section 4.3, Table 2(a)] The 'Fairness of Using Image Editing Data for Pre-Training' ablation is the right control, but the text only says SPRC was pre-trained 'under the same dataset and training strategy.' It is not stated whether the same target-benchmark Stage-1 checkpoint selection was applied to SPRC. If it was not, the control does not rule out the selection artifact identified above. Please specify the exact protocol used for the SPRC-pretrained baseline; if the selection was applied, the numbers should be presented with the same detail.
  3. [Tables 1 and 3] All reported results are from a single run with no standard deviations or confidence intervals. This is particularly concerning because some margins over strong baselines are small: on CIRR R@1, DetailFusion(Ours) is 54.55 versus SPRC† at 55.06, and on FashionIQ the average is 66.50 versus 66.41 for SPRC†. The 'outperforms previous methods' claim would be much more robust with repeated runs or, at minimum, an explicit statement of training variance. I would like to see either error bars over at least 3-5 seeds for the main comparisons, or a clear explanation of why such small differences should be treated as reliable.
minor comments (5)
  1. [Equation (2)] The denominator appears to contain a typo: 'L' is used where a plus sign is intended in the sum over S(D(Q(i)), D(I_t(j))) and S(D(Q(i)), D(I_r(j))). The definition 'S(A, BL C) := S(A,B) + S(A,C)' should be rewritten with an explicit plus sign and a short explanation.
  2. [Equation (6)] Equation (6) uses identical notation D(Q) = G(Q) = LinearH(EH(EI(Ir), Tm)), but the DI and GM branches have separate parameters according to Figure 3(a). Please use distinct symbols for the two branches' mapping and encoder parameters so that the shared-architecture but non-shared-parameter setup is unambiguous.
  3. [Section 3.4] The text says 'the DI branch also serves as the image encoder,' while Figure 3(a) says the image encoder shares parameters with the DI branch. Please clarify whether the image encoder is a separate frozen module or is literally the image side of the DI branch; this affects how Equations (6)-(7) should be interpreted.
  4. [Table 3 and Appendix B] The row 'SPRC2* [24]' appears in Table 3 without a definition or a reference entry. If this is a variant from [24], it should be described in the table caption or in Appendix B.
  5. [Abstract] The GitHub URL is given in the abstract and the conclusion promises public code. For a journal submission, please ensure that the repository is actually public and contains the training and evaluation code before acceptance; otherwise remove the URL from the abstract.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the paper's claims are empirical benchmark results against external baselines; the checkpoint-selection protocol is a benchmarking concern, not a circular derivation.

full rationale

DetailFusion is an empirical systems paper: the central claims are benchmark numbers on FashionIQ and CIRR obtained by supervised training and evaluated against external baselines. The Detail-oriented Inference branch is pretrained on the IPr2Pr editing dataset with the contrastive loss in Eq. (2), then jointly fine-tuned with the global branch via Eqs. (3)-(4), and the Compositor is trained from scratch with Eq. (5). None of these objectives defines a reported quantity in terms of the metric it is used to support; the retrieval losses are standard contrastive formulations whose target is the same retrieval task, and the gains are measured on held-out test or validation partitions relative to published methods. The one noteworthy protocol element is Sec. 4.1's statement that 'the best-performing iteration on CIRR and FashionIQ is selected as the initialization for the next stage.' Because FashionIQ results are reported on the validation set, this creates a potential model-selection or evaluation-fairness concern rather than a circularity: no fitted parameter is renamed as a prediction, no load-bearing argument reduces by construction to its own input, and no self-citation chain is used to justify the central result. The ablations also include a controlled comparison where SPRC is pretrained under the same data and strategy, so the detail-enhancement claim is not definitionally forced. Overall, the paper is self-contained as an empirical contribution and does not exhibit the circularity patterns defined for this analysis.

Assumptions & free parameters 4 free parameters · 4 assumptions · 0 invented entities

The framework's empirical gains rest on pretrained backbones, a synthetic editing dataset, and a handful of validation-tuned hyperparameters. None of these are derived from first principles, but they are standard practice for supervised CIR. The most consequential choices are the trade-off gamma and the checkpoint selection, both tuned on the target validation sets.

free parameters (4)
  • loss trade-off gamma = 2.0 (CIRR), 1.5 (FashionIQ)
    Chosen by validation performance in Fig. 4 and Tables 13-14; the central claim of optimal fusion depends on this hyperparameter.
  • temperature tau = 0.07
    Standard CLIP temperature; set in Implementation Details, affects contrastive loss sharpness.
  • cross-attention layers M,N = M=N=2
    Selected via ablation in Appendix C.3, Table 8.
  • Stage 1 checkpoint selection = best iteration on CIRR/FashionIQ
    The pre-trained DI branch initialization is chosen by downstream validation performance, a model selection step.
assumptions (4)
  • domain assumption BLIP-2 Q-Former and EVA-CLIP ViT-G/14 provide strong, transferable multimodal representations.
    Used as frozen/initialized backbones for all branches (Section 4.1).
  • domain assumption The InstructPix2Pix (IPr2Pr) dataset provides atomic detail variation priors with high reference-target similarity and single-step text edits.
    Relied on in Stage 1 pretraining (Section 3.3).
  • domain assumption Batch-based contrastive learning with reference images as hard negatives induces detail-focused features without sacrificing global semantics.
    Central to the DI loss in Eq. 2; validated only by the paper's own ablations.
  • domain assumption The evaluation metrics (Recall@K, Recallsubset@K) and benchmark splits accurately reflect CIR performance.
    Standard in the field; used throughout Section 4.

how reviews work

0 comments
Cite this review

Pith. "Pith review of DetailFusion: A Dual-branch Framework with Detail Enhancement for Composed Image Retrieval." pith.science (2026). https://pith.science/paper/PZ6CHAJN

@misc{pith2026250517796,
  author       = {Pith},
  title        = {Pith review of: DetailFusion: A Dual-branch Framework with Detail Enhancement for Composed Image Retrieval},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/PZ6CHAJN}},
  note         = {Machine review of arXiv:2505.17796}
}
read the original abstract

Composed Image Retrieval (CIR) aims to retrieve target images from a gallery based on a reference image and modification text as a combined query. Recent approaches focus on balancing global information from two modalities and encode the query into a unified feature for retrieval. However, due to insufficient attention to fine-grained details, these coarse fusion methods often struggle with handling subtle visual alterations or intricate textual instructions. In this work, we propose DetailFusion, a novel dual-branch framework that effectively coordinates information across global and detailed granularities, thereby enabling detail-enhanced CIR. Our approach leverages atomic detail variation priors derived from an image editing dataset, supplemented by a detail-oriented optimization strategy to develop a Detail-oriented Inference Branch. Furthermore, we design an Adaptive Feature Compositor that dynamically fuses global and detailed features based on fine-grained information of each unique multimodal query. Extensive experiments and ablation analyses not only demonstrate that our method achieves state-of-the-art performance on both CIRR and FashionIQ datasets but also validate the effectiveness and cross-domain adaptability of detail enhancement for CIR.

Figures

Figures reproduced from arXiv: 2505.17796 by the authors.

Figure 1
Figure 1. Workflows of existing supervised CIR methods and Ours: (a) Late fusion, (b) Textual [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Overall pipeline of our proposed DetailFusion. The upper section illustrates the training phase, while the lower-left displays the datasets utilized at each training stage, and the lower-right illustrates the inference phase. During the training phase, the DI branch is first pre-trained on the IPr2Pr dataset [11], followed by joint fine-tuning of both the DI and GM branches on CIR datasets. Finally, the parameters o… view at source ↗
Figure 3
Figure 3. Module architecture of our proposed DetailFusion. (a) Illustration of the shared structure between the DI and GM branches. A hybrid-modal encoder separately encodes the query and image, corresponding to the left and right branches, respectively. The image encoder shares parameters with the DI branch, but non-shared with the GM branch. (b) Illustration of the Adaptive Feature Compositor. Features from both branches f… view at source ↗
Figures from the paper (3 more)
Figure 4
Figure 4. Figure 4: Analysis of different values of trade￾off hyper-parameter γ during joint fine-tuning. that the image editing dataset, due to its distribu￾tional deviation, is not suitable for direct CIR pre￾training. However, by incorporating this dataset into our method, we effective…
Figure 5
Figure 5. Figure 5: Qualitative comparison of our method and the Baseline on the CIRR validation set. Images are arranged in descending order from left to right based on similarity to the multimodal query. The green boxes highlight the target image, while all non-target images are marked …
Figure 6
Figure 6. Figure 6: Qualitative comparison of our method and SPRC on the CIRR validation set. Images are arranged in descending order from left to right based on similarity to the multimodal query. The green boxes highlight the target image, while all non-target images are marked with red…

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

65 extracted references · 46 canonical work pages

  1. [1]

    Composing text and image for image retrieval-an empirical odyssey

    Nam V o, Lu Jiang, Chen Sun, Kevin Murphy, Li-Jia Li, Li Fei-Fei, and James Hays. Composing text and image for image retrieval-an empirical odyssey. InProceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 6439–6448, 2019

    2019

  2. [2]

    Image retrieval on real-life images with pre-trained vision-and-language models

    Zheyuan Liu, Cristian Rodriguez-Opazo, Damien Teney, and Stephen Gould. Image retrieval on real-life images with pre-trained vision-and-language models. InProceedings of the IEEE/CVF International Conference on Computer Vision, pages 2125–2134, 2021

    2021

  3. [3]

    Deep image retrieval: Learning global representations for image search

    Albert Gordo, Jon Almazán, Jerome Revaud, and Diane Larlus. Deep image retrieval: Learning global representations for image search. InComputer Vision–ECCV 2016: 14th European Conference, Amsterdam, The Netherlands, October 11-14, 2016, Proceedings, Part VI 14, pages 241–257. Springer, 2016

    work page 2016

  4. [4]

    Large-scale image retrieval with attentive deep local features

    Hyeonwoo Noh, Andre Araujo, Jack Sim, Tobias Weyand, and Bohyung Han. Large-scale image retrieval with attentive deep local features. InProceedings of the IEEE international conference on computer vision, pages 3456–3465, 2017

    work page 2017

  5. [5]

    Camp: Cross- modal adaptive message passing for text-image retrieval

    Zihao Wang, Xihui Liu, Hongsheng Li, Lu Sheng, Junjie Yan, Xiaogang Wang, and Jing Shao. Camp: Cross- modal adaptive message passing for text-image retrieval. InProceedings of the IEEE/CVF international conference on computer vision, pages 5764–5773, 2019

    work page 2019

  6. [6]

    Context-aware attention network for image-text retrieval

    Qi Zhang, Zhen Lei, Zhaoxiang Zhang, and Stan Z Li. Context-aware attention network for image-text retrieval. InProceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 3536–3545, 2020

    work page 2020

  7. [7]

    Image search with text feedback by visiolinguistic atten- tion learning

    Yanbei Chen, Shaogang Gong, and Loris Bazzani. Image search with text feedback by visiolinguistic atten- tion learning. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 3001–3011, 2020

    work page 2020

  8. [8]

    Learning transferable visual models from natural language supervision

    Alec Radford, Jong Wook Kim, Chris Hallacy, Aditya Ramesh, Gabriel Goh, Sandhini Agarwal, Girish Sastry, Amanda Askell, Pamela Mishkin, Jack Clark, et al. Learning transferable visual models from natural language supervision. InInternational conference on machine learning, pages 8748–8763. PMLR, 2021

    2021

Show all 65 references
  1. [9]

    Blip: Bootstrapping language-image pre-training for unified vision-language understanding and generation

    Junnan Li, Dongxu Li, Caiming Xiong, and Steven Hoi. Blip: Bootstrapping language-image pre-training for unified vision-language understanding and generation. InInternational conference on machine learning, pages 12888–12900. PMLR, 2022

    2022

  2. [10]

    Blip-2: Bootstrapping language-image pre-training with frozen image encoders and large language models

    Junnan Li, Dongxu Li, Silvio Savarese, and Steven Hoi. Blip-2: Bootstrapping language-image pre-training with frozen image encoders and large language models. InInternational conference on machine learning, pages 19730–19742. PMLR, 2023

    2023

  3. [11]

    Instructpix2pix: Learning to follow image editing instructions

    Tim Brooks, Aleksander Holynski, and Alexei A Efros. Instructpix2pix: Learning to follow image editing instructions. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 18392–18402, 2023

    2023

  4. [12]

    Artemis: Attention-based re- trieval with text-explicit matching and implicit similarity

    Ginger Delmas, Rafael S Rezende, Gabriela Csurka, and Diane Larlus. Artemis: Attention-based re- trieval with text-explicit matching and implicit similarity. InInternational Conference on Learning Representations, 2022

    2022

  5. [13]

    Compositional learning of image-text query for image retrieval

    Muhammad Umer Anwaar, Egor Labintcev, and Martin Kleinsteuber. Compositional learning of image-text query for image retrieval. InProceedings of the IEEE/CVF Winter conference on Applications of Computer Vision, pages 1140–1149, 2021. 10

    2021

  6. [14]

    Alberto Baldrati, Marco Bertini, Tiberio Uricchio, and Alberto Del Bimbo. Composed image retrieval using contrastive learning and task-oriented clip-based features.ACM Transactions on Multimedia Computing, Communications and Applications, 20(3):1–24, 2023

    2023

  7. [15]

    Spirit: style-guided patch interaction for fashion im- age retrieval with text feedback.ACM Transactions on Multimedia Computing, Communications and Applications, 20(6):1–17, 2024

    Yanzhe Chen, Jiahuan Zhou, and Yuxin Peng. Spirit: style-guided patch interaction for fashion im- age retrieval with text feedback.ACM Transactions on Multimedia Computing, Communications and Applications, 20(6):1–17, 2024

    2024

  8. [16]

    Shenshen Li, Xing Xu, Xun Jiang, Fumin Shen, Zhe Sun, and Andrzej Cichocki. Cross-modal attention preservation with self-contrastive learning for composed query-based image retrieval.ACM Transactions on Multimedia Computing, Communications and Applications, 20(6):1–22, 2024

    2024

  9. [17]

    Data roaming and quality assessment for composed image retrieval.Proceedings of the AAAI Conference on Artificial Intelligence, 38(4):2991–2999, Mar

    Matan Levy, Rami Ben-Ari, Nir Darshan, and Dani Lischinski. Data roaming and quality assessment for composed image retrieval.Proceedings of the AAAI Conference on Artificial Intelligence, 38(4):2991–2999, Mar. 2024

    2024

  10. [18]

    Candidate set re-ranking for composed image retrieval with dual multi-modal encoder.Transactions on Machine Learning Research, 2024

    Zheyuan Liu, Weixuan Sun, Damien Teney, and Stephen Gould. Candidate set re-ranking for composed image retrieval with dual multi-modal encoder.Transactions on Machine Learning Research, 2024

    2024

  11. [19]

    Simple but effective raw-data level multimodal fusion for composed image retrieval

    Haokun Wen, Xuemeng Song, Xiaolin Chen, Yinwei Wei, Liqiang Nie, and Tat-Seng Chua. Simple but effective raw-data level multimodal fusion for composed image retrieval. InProceedings of the 47th International ACM SIGIR Conference on Research and Development in Information Retri...

    2024

  12. [20]

    this is my unicorn, fluffy

    Niv Cohen, Rinon Gal, Eli A Meirom, Gal Chechik, and Yuval Atzmon. “this is my unicorn, fluffy”: Personalizing frozen vision-language representations. InEuropean conference on computer vision, pages 558–577. Springer, 2022

    2022

  13. [21]

    Effective conditioned and composed image retrieval combining clip-based features

    Alberto Baldrati, Marco Bertini, Tiberio Uricchio, and Alberto Del Bimbo. Effective conditioned and composed image retrieval combining clip-based features. InProceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 21466–21474, 2022

    2022

  14. [22]

    Pic2word: Mapping pictures to words for zero-shot composed image retrieval

    Kuniaki Saito, Kihyuk Sohn, Xiang Zhang, Chun-Liang Li, Chen-Yu Lee, Kate Saenko, and Tomas Pfister. Pic2word: Mapping pictures to words for zero-shot composed image retrieval. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 19305–19314, 2023

    2023

  15. [23]

    Zero-shot composed image retrieval with textual inversion

    Alberto Baldrati, Lorenzo Agnolucci, Marco Bertini, and Alberto Del Bimbo. Zero-shot composed image retrieval with textual inversion. InProceedings of the IEEE/CVF International Conference on Computer Vision, pages 15338–15347, 2023

    2023

  16. [24]

    Sentence-level prompts benefit composed image retrieval

    Yang Bai, Xinxing Xu, Yong Liu, Salman Khan, Fahad Khan, Wangmeng Zuo, Rick Siow Mong Goh, and Chun-Mei Feng. Sentence-level prompts benefit composed image retrieval. InThe Twelfth International Conference on Learning Representations, 2024

    2024

  17. [25]

    Covr: Learning composed video retrieval from web video captions.Proceedings of the AAAI Conference on Artificial Intelligence, 38(6):5270–5279, March 2024

    Lucas Ventura, Antoine Yang, Cordelia Schmid, and Gül Varol. Covr: Learning composed video retrieval from web video captions.Proceedings of the AAAI Conference on Artificial Intelligence, 38(6):5270–5279, March 2024

    2024

  18. [26]

    MagicLens: Self-supervised image retrieval with open-ended instructions

    Kai Zhang, Yi Luan, Hexiang Hu, Kenton Lee, Siyuan Qiao, Wenhu Chen, Yu Su, and Ming-Wei Chang. MagicLens: Self-supervised image retrieval with open-ended instructions. In Ruslan Salakhutdinov, Zico Kolter, Katherine Heller, Adrian Weller, Nuria Oliver, Jonathan Scarlett, and ...

    2024

  19. [27]

    Making the v in vqa matter: Elevating the role of image understanding in visual question answering

    Yash Goyal, Tejas Khot, Aishwarya Agrawal, Douglas Summers-Stay, Dhruv Batra, and Devi Parikh. Making the v in vqa matter: Elevating the role of image understanding in visual question answering. International Journal of Computer Vision, page 398–414, Apr 2019

    2019

  20. [28]

    Frozen in time: A joint video and image encoder for end-to-end retrieval

    Max Bain, Arsha Nagrani, Gül Varol, and Andrew Zisserman. Frozen in time: A joint video and image encoder for end-to-end retrieval. InProceedings of the IEEE/CVF international conference on computer vision, pages 1728–1738, 2021

    2021

  21. [29]

    Llama: Open and efficient foundation language models.arXiv preprint arXiv:2302.13971, 2023

    Hugo Touvron, Thibaut Lavril, Gautier Izacard, Xavier Martinet, Marie-Anne Lachaux, Timothée Lacroix, Baptiste Rozière, Naman Goyal, Eric Hambro, Faisal Azhar, et al. Llama: Open and efficient foundation language models.arXiv preprint arXiv:2302.13971, 2023. 11

    2023 arXiv

  22. [30]

    Tom B. Brown, Benjamin Mann, Nick Ryder, Melanie Subbiah, Jared Kaplan, Prafulla Dhariwal, Arvind Neelakantan, Pranav Shyam, Girish Sastry, Amanda Askell, Sandhini Agarwal, Ariel Herbert-V oss, Gretchen Krueger, Tom Henighan, Rewon Child, Aditya Ramesh, Daniel M. Ziegler, Jeff...

    2005 arXiv

  23. [31]

    Compodiff: Versatile composed image retrieval with latent diffusion.Transactions on Machine Learning Research,

    Geonmo Gu, Sanghyuk Chun, Wonjae Kim, HeeJae Jun, Yoohoon Kang, and Sangdoo Yun. Compodiff: Versatile composed image retrieval with latent diffusion.Transactions on Machine Learning Research,

  24. [32]

    High-resolution image synthesis with latent diffusion models

    Robin Rombach, Andreas Blattmann, Dominik Lorenz, Patrick Esser, and Bjorn Ommer. High-resolution image synthesis with latent diffusion models. In2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), Jun 2022

    2022

  25. [33]

    Fashion iq: A new dataset towards retrieving images by natural language feedback

    Hui Wu, Yupeng Gao, Xiaoxiao Guo, Ziad Al-Halah, Steven Rennie, Kristen Grauman, and Rogerio Feris. Fashion iq: A new dataset towards retrieving images by natural language feedback. In2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 11302–11312, 2021

    2021

  26. [34]

    A corpus for reasoning about natural language grounded in photographs

    Alane Suhr, Stephanie Zhou, Ally Zhang, Iris Zhang, Huajun Bai, and Yoav Artzi. A corpus for reasoning about natural language grounded in photographs. InProceedings of the 57th Annual Meeting of the Association for Computational Linguistics, Jan 2019

    2019

  27. [35]

    Eva: Exploring the limits of masked visual representation learning at scale

    Yuxin Fang, Wen Wang, Binhui Xie, Quan Sun, Ledell Wu, Xinggang Wang, Tiejun Huang, Xinlong Wang, and Yue Cao. Eva: Exploring the limits of masked visual representation learning at scale. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, page...

    2023

  28. [36]

    Decoupled weight decay regularization.arXiv preprint arXiv:1711.05101, 2017

    I Loshchilov. Decoupled weight decay regularization.arXiv preprint arXiv:1711.05101, 2017

    2017 arXiv

  29. [37]

    Improving composed image retrieval via contrastive learning with scaling positives and negatives

    Zhangchi Feng, Richong Zhang, and Zhijie Nie. Improving composed image retrieval via contrastive learning with scaling positives and negatives. InProceedings of the 32nd ACM International Conference on Multimedia, pages 1632–1641, 2024

    2024

  30. [38]

    Decomposing semantic shifts for composed image retrieval

    Xingyu Yang, Daqing Liu, Heng Zhang, Yong Luo, Chaoyue Wang, and Jing Zhang. Decomposing semantic shifts for composed image retrieval. InProceedings of the AAAI Conference on Artificial Intelligence, volume 38, pages 6576–6584, 2024

    2024

  31. [39]

    Bi-directional training for composed image retrieval via text prompt learning

    Zheyuan Liu, Weixuan Sun, Yicong Hong, Damien Teney, and Stephen Gould. Bi-directional training for composed image retrieval via text prompt learning. InProceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision, pages 5753–5762, 2024

    2024

  32. [40]

    Cala: Com- plementary association learning for augmenting comoposed image retrieval

    Xintong Jiang, Yaxiong Wang, Mengjian Li, Yujiao Wu, Bingwen Hu, and Xueming Qian. Cala: Com- plementary association learning for augmenting comoposed image retrieval. InProceedings of the 47th International ACM SIGIR Conference on Research and Development in Information Retri...

    2024

  33. [41]

    Target-guided composed image retrieval

    Haokun Wen, Xian Zhang, Xuemeng Song, Yinwei Wei, and Liqiang Nie. Target-guided composed image retrieval. InProceedings of the 31st ACM International Conference on Multimedia, pages 915–923, 2023

    2023

  34. [42]

    Leveraging large vision-language model as user intent-aware encoder for composed image retrieval

    Zelong Sun, Dong Jing, Guoxing Yang, Nanyi Fei, and Zhiwu Lu. Leveraging large vision-language model as user intent-aware encoder for composed image retrieval. InProceedings of the AAAI Conference on Artificial Intelligence, volume 39, pages 7149–7157, 2025

    2025

  35. [43]

    Ccin: Compositional conflict identification and neutralization for composed image retrieval

    Likai Tian, Jian Zhao, Zechao Hu, Zhengwei Yang, Hao Li, Lei Jin, Zheng Wang, and Xuelong Li. Ccin: Compositional conflict identification and neutralization for composed image retrieval. 2025

    2025

  36. [44]

    Conditioned and composed image retrieval combining and partially fine-tuning clip-based features

    Alberto Baldrati, Marco Bertini, Tiberio Uricchio, and Alberto Del Bimbo. Conditioned and composed image retrieval combining and partially fine-tuning clip-based features. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 4959–4968, 2022

    2022

  37. [45]

    Few-shot composition learning for image retrieval with prompt tuning

    Junda Wu, Rui Wang, Handong Zhao, Ruiyi Zhang, Chaochao Lu, Shuai Li, and Ricardo Henao. Few-shot composition learning for image retrieval with prompt tuning. InProceedings of the AAAI Conference on Artificial Intelligence, volume 37, pages 4729–4737, 2023

    2023

  38. [46]

    Multi-modal transformer with global-local alignment for composed query image retrieval.IEEE Transactions on Multimedia, 25:8346–8357, 2023

    Yahui Xu, Yi Bin, Jiwei Wei, Yang Yang, Guoqing Wang, and Heng Tao Shen. Multi-modal transformer with global-local alignment for composed query image retrieval.IEEE Transactions on Multimedia, 25:8346–8357, 2023. 12

    2023

  39. [47]

    Negative-sensitive framework with semantic enhancement for composed image retrieval.IEEE Transactions on Multimedia, 26:7608–7621, 2024

    Yifan Wang, Liyuan Liu, Chun Yuan, Minbo Li, and Jing Liu. Negative-sensitive framework with semantic enhancement for composed image retrieval.IEEE Transactions on Multimedia, 26:7608–7621, 2024

    2024

  40. [48]

    Collaborative group: Composed image retrieval via consensus learning from noisy annotations.Know.-Based Syst., 300(C), November 2024

    Xu Zhang, Zhedong Zheng, Linchao Zhu, and Yi Yang. Collaborative group: Composed image retrieval via consensus learning from noisy annotations.Know.-Based Syst., 300(C), November 2024

    2024

  41. [49]

    Pseudo-triplet guided few-shot composed image retrieval.arXiv preprint arXiv:2407.06001, 2024

    Bohan Hou, Haoqiang Lin, Haokun Wen, Meng Liu, Mingzhu Xu, and Xuemeng Song. Pseudo-triplet guided few-shot composed image retrieval.arXiv preprint arXiv:2407.06001, 2024

    2024 arXiv

  42. [50]

    Vqa4cir: Boosting composed image retrieval with visual question answering

    Chun-Mei Feng, Yang Bai, Tao Luo, Zhen Li, Salman Khan, Wangmeng Zuo, Rick Siow Mong Goh, and Yong Liu. Vqa4cir: Boosting composed image retrieval with visual question answering. InProceedings of the AAAI Conference on Artificial Intelligence, volume 39, pages 2942–2950, 2025

    2025

  43. [51]

    Finecir: Explicit parsing of fine-grained modification semantics for composed image retrieval.arXiv preprint arXiv:2503.21309, 2025

    Zixu Li, Zhiheng Fu, Yupeng Hu, Zhiwei Chen, Haokun Wen, and Liqiang Nie. Finecir: Explicit parsing of fine-grained modification semantics for composed image retrieval.arXiv preprint arXiv:2503.21309, 2025

    2025 arXiv

  44. [52]

    An image is worth one word: Personalizing text-to-image generation using textual inversion

    Rinon Gal, Yuval Alaluf, Yuval Atzmon, Or Patashnik, Amit H Bermano, Gal Chechik, and Daniel Cohen-Or. An image is worth one word: Personalizing text-to-image generation using textual inversion. arXiv preprint arXiv:2208.01618, 2022

    2022 arXiv

  45. [53]

    Knowledge-enhanced dual-stream zero-shot composed image retrieval

    Yucheng Suo, Fan Ma, Linchao Zhu, and Yi Yang. Knowledge-enhanced dual-stream zero-shot composed image retrieval. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 26951–26962, 2024

    2024

  46. [54]

    Ldre: Llm-based divergent reasoning and ensemble for zero-shot composed image retrieval

    Zhenyu Yang, Dizhan Xue, Shengsheng Qian, Weiming Dong, and Changsheng Xu. Ldre: Llm-based divergent reasoning and ensemble for zero-shot composed image retrieval. InProceedings of the 47th International ACM SIGIR Conference on Research and Development in Information Retrieval...

    2024

  47. [55]

    Semantic editing increment benefits zero-shot composed image retrieval

    Zhenyu Yang, Shengsheng Qian, Dizhan Xue, Jiahong Wu, Fan Yang, Weiming Dong, and Changsheng Xu. Semantic editing increment benefits zero-shot composed image retrieval. InProceedings of the 32nd ACM International Conference on Multimedia, pages 1245–1254, 2024

    2024

  48. [56]

    Vision-by-language for training-free compositional image retrieval.arXiv preprint arXiv:2310.09291, 2023

    Shyamgopal Karthik, Karsten Roth, Massimiliano Mancini, and Zeynep Akata. Vision-by-language for training-free compositional image retrieval.arXiv preprint arXiv:2310.09291, 2023

    2023 arXiv

  49. [57]

    A comprehensive survey on composed image retrieval.arXiv preprint arXiv:2502.18495, 2025

    Xuemeng Song, Haoqiang Lin, Haokun Wen, Bohan Hou, Mingzhu Xu, and Liqiang Nie. A comprehensive survey on composed image retrieval.arXiv preprint arXiv:2502.18495, 2025

    2025 arXiv

  50. [58]

    Modality-agnostic attention fusion for visual search with text feedback.arXiv preprint arXiv:2007.00145, 2020

    Eric Dodds, Jack Culpepper, Simao Herdade, Yang Zhang, and Kofi Boakye. Modality-agnostic attention fusion for visual search with text feedback.arXiv preprint arXiv:2007.00145, 2020

    2007 arXiv

  51. [59]

    Cosmo: Content-style modulation for image retrieval with text feedback

    Seungmin Lee, Dongwan Kim, and Bohyung Han. Cosmo: Content-style modulation for image retrieval with text feedback. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 802–812, 2021

    2021

  52. [60]

    Fame-vil: Multi-tasking vision-language model for heterogeneous fashion tasks

    Xiao Han, Xiatian Zhu, Licheng Yu, Li Zhang, Yi-Zhe Song, and Tao Xiang. Fame-vil: Multi-tasking vision-language model for heterogeneous fashion tasks. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 2669–2680, 2023

    2023

  53. [61]

    Composed image retrieval with text feedback via multi-grained uncertainty regularization

    Yiyang Chen, Zhedong Zheng, Wei Ji, Leigang Qu, and Tat-Seng Chua. Composed image retrieval with text feedback via multi-grained uncertainty regularization. InInternational Conference on Learning Representations (ICLR), 2024

    2024

  54. [62]

    Genecis: A benchmark for general conditional image similarity

    Sagar Vaze, Nicolas Carion, and Ishan Misra. Genecis: A benchmark for general conditional image similarity. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 6862–6872, 2023

    2023

  55. [63]

    Language-only training of zero-shot composed image retrieval

    Geonmo Gu, Sanghyuk Chun, Wonjae Kim, Yoohoon Kang, and Sangdoo Yun. Language-only training of zero-shot composed image retrieval. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 13225–13234, 2024

    2024

  56. [64]

    Missing target-relevant information prediction with world model for accurate zero-shot composed image retrieval

    Yuanmin Tang, Jing Yu, Keke Gai, Jiamin Zhuang, Gang Xiong, Gaopeng Gou, and Qi Wu. Missing target-relevant information prediction with world model for accurate zero-shot composed image retrieval. arXiv preprint arXiv:2503.17109, 2025. 13 A Scaling Positives and Negatives for ...

    2025 arXiv

  57. [2024]

    Expert Certification

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.