Pith. sign in

REVIEW 3 major objections 6 minor 88 references

Visual identity discrimination, long missing from universal multimodal embeddings, can be added through identity-aware sampling and a unified contrastive loss, yielding ~25-point gains on a new MVEB benchmark without hurting general perform

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

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

T0 review

2026-08-04 20:37 UTC pith:425HBYSY

load-bearing objection A genuinely useful benchmark and training recipe for identity-aware UMEs, but the undisclosed random-split fallback muddies the out-of-domain claim and needs to be addressed before I'd trust the numbers. the 3 major comments →

arxiv 2608.01794 v1 pith:425HBYSY submitted 2026-08-03 cs.CV cs.AIcs.CL

Illuminating Visual Identity in Universal Multimodal Embeddings

Jiawei Cao , Junyi Feng , Jiashen Hua , Ziheng Huang , Bing Deng , Kaijie Wu , Chaochen Gu , Jieping Ye This is my paper
classification cs.CV cs.AIcs.CL
keywords visual identity discriminationuniversal multimodal embeddingscontrastive learningidentity-aware samplingmultimodal retrievalbenchmarkre-identificationidentity-preserving generation
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

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

This paper argues that universal multimodal embedding (UME) models, which align images and text in one space, have silently neglected visual identity discrimination: telling whether two images show the same person, product, landmark, or edited subject. The authors formalize this as VisID, split into four meta-tasks (recognition, re-identification, grounding, editing), and build MVEB, a 28-dataset benchmark with 522K samples to train and test it. They then show that a simple add-on—offline identity-aware batch sampling plus structured hard negatives under one contrastive loss, called VisME—raises MVEB scores by roughly 25–30 points over the best prior UMEs while matching or slightly beating them on the general MMEB benchmark. If correct, this means identity-level retrieval can be a first-class capability of general-purpose multimodal embeddings, with direct use in instance retrieval, person/vehicle re-ID, and identity-preserving AI generation.

Core claim

Visual identity discrimination—matching a query image to other instances of the same person, product, landmark, or edited subject—is treated here as a missing core capability of universal multimodal embeddings. The authors define four meta-tasks (recognition, re-identification, grounding, editing), build MVEB (28 datasets, 522K samples, identity-disjoint splits where feasible), and train VisME with offline identity-aware batch pre-scheduling, structured hard negatives, and a single contrastive loss. VisME's MVEB averages of 69.1 (2B) and 78.8 (7B) exceed the strongest prior UME by roughly 25–30 points while matching MMEB (72.2 vs 72.0), supporting the claim that identity and generality can b

What carries the argument

The load-bearing mechanism is offline identity-aware batch pre-scheduling: all mini-batches are generated before training so each identity appears at most once per batch, with sampling probability proportional to identity size, eliminating false negatives from repeated identities. Structured hard-negative sampling then inserts designated same-semantics/different-identity pairs, and a single unified contrastive loss with scaled cosine similarity (learnable temperature) optimizes both semantic and identity triplets. The MVEB curation pipeline, including identity-disjoint train/test splits, is the measurement machinery that makes the capability visible.

Load-bearing premise

The formulation equates identity with the label each source dataset provides, so across datasets identity ranges from a unique instance (a face, a product) to a fine-grained category (a species, a landmark), and for datasets with too few identities the evaluation split is random rather than identity-disjoint, meaning some test identities may already be in training.

What would settle it

Train and evaluate on a version of MVEB where identity is defined strictly at instance level (each identity is one unique object/person/face, not a category), and where every sub-dataset uses a fully identity-disjoint split with no identity overlap between train and test. If the ~25-point advantage of VisME over baselines shrinks substantially or vanishes, the central claim would be shown to depend on category-level labels and identity leakage rather than on true identity discrimination.

Watch this falsifier. Get emailed when new claim-graph text bears on it.

Share X LinkedIn Reddit HN

If this is right

  • UME models trained with the VisME recipe gain a capability that previous universal embeddings lacked: matching a query image to other instances of the same identity under a language instruction.
  • MVEB supplies a reusable 28-dataset evaluation across four identity meta-tasks, with 8 out-of-domain sets, so future models can be compared on identity discrimination directly.
  • Identity-aware embeddings of this kind can serve as data-curation tools and evaluation metrics for identity-preserving AI-generated content.
  • Because identity-aware sampling removes false negatives, training is more stable at large batch sizes and benefits from hard-negative mining, which the ablations show improves both in-distribution and out-of-distribution scores.
  • Scaling the base MLLM from 3B to 7B parameters consistently improves identity performance, with the largest gains on face datasets, suggesting that larger models capture finer identity cues.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • If the identity-aware batch pre-scheduling generalizes, the same false-negative-avoidance trick could improve contrastive learning in other settings where the same entity reappears across samples, such as audio speaker embeddings or document embeddings with repeated entities.
  • MVEB's 'identity' includes fine-grained categories (species, landmarks, product types), so part of the reported gain may be improved fine-grained classification rather than true instance-level discrimination; a strict instance-only split would reveal the size of that gap.
  • The paper's reliance on an auxiliary embedding model for hard-negative mining suggests a self-improving loop: better identity embeddings would produce better hard negatives, which could in turn train even better identity embeddings.
  • The supplementary candidate-pool scaling experiment shows that even VisME degrades when the gallery grows from 3k to 100k candidates, so gallery-scale identity retrieval remains an open problem that the planned larger MVEB will measure.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 6 minor

Summary. The paper argues that universal multimodal embedding (UME) models neglect visual identity discrimination (VisID) and proposes to close this gap in three parts: a formal decomposition of VisID into four meta-tasks (Identity Recognition, Re-Identification, Identity Grounding, Identity Editing); a new benchmark, MVEB, curated from 28 real and synthetic datasets, with 20 training and 8 out-of-domain (OOD) evaluation sets; and a training framework, VisME, that augments standard UME contrastive training with identity-aware offline sampling and structured hard negatives. Experiments on Qwen2-VL/Qwen2.5-VL backbones show large MVEB gains over existing UMEs (e.g., MVEB average 78.8 for the 7B model vs. 52.3 for VLM2Vec-7B) while roughly matching MMEB performance. Code and data are promised for release.

Significance. If the results hold, this is a useful contribution. VisID is indeed underexplored in the UME literature, and the proposed four-task decomposition is a reasonable organizing principle. MVEB, despite being author-constructed, is larger and more identity-focused than existing UME benchmarks and would be a valuable community resource if released with clean split metadata. The method itself is simple and compatible with existing UME pipelines, and the reported gains on identity-centric tasks are substantial. The external MMEB evaluation and the OOD splits provide some check against pure overfitting to the proposed benchmark. However, the strength of the central claim is weakened by a self-constructed benchmark with an admitted identity-leakage fallback, by single-run comparisons with no error bars, and by a definition of 'identity' that mixes instance-level and category-level labels.

major comments (3)
  1. [Sec. 4.1 Step 3, Table 6, Supp. C] The benchmark's identity-aware split has an exception that directly affects the headline numbers. The text states: 'Crucially, whenever feasible, we enforce a strict identity-aware split... For sub-datasets with insufficient unique identities, we fall back to a standard random split at the sample level.' Table 6 marks Cars196 and iNat as 'ID-Unseen? = ✗', and both are also training sub-datasets (Figure 3). Thus the reported 'All IND' and overall MVEB averages include in-domain evaluations on identities that may have appeared in training. The supplementary statement that 'there is no overlap of specific images between the training and evaluation sets' does not address identity overlap. I acknowledge that the eight OOD rows are all marked 'ID-Unseen? = ✔', so the OOD average may be less affected; the problem is the overall/IND scores and the general claim of generalizing to unseen identiti
  2. [Tables 1, 2, 4, 5] All experiments appear to be single runs with no error bars or significance tests. This matters most for the MMEB claim: VisME (Qwen2.5-VL-7B) scores 72.2 vs. B3's 72.0, a 0.2-point difference, and the text states 'Our method establishes best results on both MMEB and MVEB benchmarks.' With a single run, a 0.2-point gap is not distinguishable from training noise. Similarly, the temperature ablations in Table 5 differ by only 0.1–0.5 points, and the interleave batch size comparison in Table 4 shows a 0.1-point difference at the top end. Please provide means and standard deviations over at least three seeds for the main results and for the comparisons where the difference is small, or perform a paired significance test. The large MVEB margins are probably robust, but the MMEB 'best' claim is not supported without this.
  3. [Sec. 3, Tables 8 and 9] The formulation defines identity as 'either a unique visual instance or a fine-grained semantic category', and the benchmark accordingly treats iNat species labels and GLDv2 landmark labels as identities alongside face and vehicle identities. This conflation is materially relevant to the paper's motivation: instance retrieval, person re-identification, and identity preservation in AIGC are about unique entities. A model that improves on 'same species' or 'same landmark' may be learning fine-grained category recognition rather than instance-level identity discrimination. Please report the MVEB results separately for instance-level sub-datasets (unique objects, faces, persons, vehicles, products) and category-level sub-datasets (iNat, GLDv2, possibly MET), and discuss whether the gains differ. This does not invalidate the benchmark, but it is necessary for interpreting what VisME actually
minor comments (6)
  1. [Sec. 4 heading] The section heading reads 'Methology'; should be 'Methodology'.
  2. [Sec. 4.2.1] There is a duplicated phrase: 'optimized under a single, consistent objective objective(q,C+,C−)'.
  3. [Eq. (4)] The notation 'C − i )' has an unbalanced parenthesis; the negative set should be written consistently as C−_i.
  4. [Sec. 5.1 / 5.2] The adaptation protocol for dual-encoder models is ad hoc: all tasks except Identity Editing are converted to pure image-to-image by dropping text, and Identity Editing uses averaged visual and text embeddings. This is not the native task format for these models and may understate their performance. Please describe the protocol more precisely, justify it, or move dual-encoder results to a clearly labeled 'adapted protocol' subsection. The central UME comparison is unaffected, but the current Table 1 mixes native and adapted evaluations.
  5. [Table 6] The 'ID-Unseen?' column is informative but is only defined in the caption. Please add a sentence in the main text explaining that this column encodes whether the identity-aware split was applied, and note explicitly in the caption that '✗' indicates the random-split fallback.
  6. [Supp. B] The baseline name is inconsistently written as 'VLM2VEC' here; elsewhere it is VLM2Vec.

Circularity Check

0 steps flagged

No circular derivation: MVEB is a self-constructed benchmark used for training and evaluation, but OOD results come from wholly withheld datasets and external MMEB results provide independent grounding.

full rationale

The paper's derivation chain is empirical rather than formal: it constructs MVEB (Sec. 4.1), trains VisME with identity-aware sampling and a unified contrastive loss (Sec. 4.2), and then measures performance on MVEB and MMEB (Sec. 5). No equation or claimed prediction reduces to an input by construction. The main self-referential element is that MVEB is both the training corpus and the evaluation benchmark; however, the 8 OOD sub-datasets (Product1m, ShopProduct, Market1501, CasiaFace, WikiPerson, FORB, IDMROOD, OpenGPT4o) are wholly absent from training, so their scores are not identity-leaked. The random-split fallback disclosed in Sec. 4.1 Step 3 ('For sub-datasets with insufficient unique identities, we fall back to a standard random split at the sample level'), together with the ✗ ID-Unseen flags in Table 6, weakens the 'unseen identity' interpretation for some in-distribution sets. The supplementary claim in Sec. C that 'there is no overlap of specific images between the training and evaluation sets' addresses image duplication, not identity overlap, so the main text's stronger 'no identity present in training appears in any evaluation set' is not fully supported for those fallback sets. This is a data-integrity caveat, not circularity: the method does not fit or rename those scores as predictions. The MVEB advantage over baselines is partly a training-data advantage (baselines were not trained on MVEB), but the paper also reports comparable MMEB numbers (72.2 vs 72.0), an external benchmark. The auxiliary embedding model [2,72,73] and hard-negative setting [55] are external, and no load-bearing self-citation or imported uniqueness theorem appears. Consequently, the central empirical claims are independently grounded and no step is equivalent to its inputs by definition.

Axiom & Free-Parameter Ledger

3 free parameters · 4 axioms · 0 invented entities

The paper's contribution is empirical: a benchmark plus a training recipe. The numeric free parameters are contrastive-loss hyperparameters ablated in Sec 5.3. The key assumptions are definitional (what counts as identity), about label quality, and about split integrity. No new physical or conceptual entities are postulated.

free parameters (3)
  • temperature tau = learnable, init 0.02
    Scale in the scaled-cosine InfoNCE loss (Eq. 3); ablated in Table 5 and found best with init 0.02.
  • hard negatives k = 5
    Number of structured hard negatives sampled per anchor; set 'same as the paper [55]' in Sec 5.1.
  • interleave batch size = 64
    Sub-dataset interleaving granularity chosen after ablation in Table 4 as the best trade-off.
axioms (4)
  • domain assumption Identity labels in source datasets are correct and unambiguous
    MVEB positives and negatives are defined entirely by dataset-provided IDs (Sec 4.1, Eq. 2). Noise in these labels propagates into both training and evaluation.
  • ad hoc to paper The four meta-tasks jointly cover visual identity discrimination
    Sec. 3 defines VisID via these meta-tasks; this is a definitional choice made by the authors, not a derived result.
  • domain assumption Auxiliary embedding model reliably selects hard negatives and denoises generated data
    Sec 4.1 Step 2 uses an auxiliary model trained following [2,72,73] to mine hard negatives and filter AI-generated data; errors in that model affect benchmark quality.
  • domain assumption Identity-aware split ensures no identity overlap between train and eval
    Sec 4.1 Step 3 claims this, but the paper admits a fallback to random sample-level splits for datasets with too few identities, so the assumption does not hold for all MVEB sub-datasets.

reviewed 2026-08-04 · how reviews work

0 comments
Cite this review

Pith. "Pith review of Illuminating Visual Identity in Universal Multimodal Embeddings." pith.science (2026). https://pith.science/paper/425HBYSY

@misc{pith2026260801794,
  author       = {Pith},
  title        = {Pith review of: Illuminating Visual Identity in Universal Multimodal Embeddings},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/425HBYSY}},
  note         = {Machine review of arXiv:2608.01794}
}
Share X LinkedIn Reddit HN
read the original abstract

Universal Multimodal Embeddings (UMEs) aim to unify various modalities and tasks into a shared representation space. In recent years, this field has witnessed substantial progress driven by the development of Multimodal Large Language Models (MLLMs). However, a crucial capability, visual identity discrimination, remains underexplored in existing UME methods, despite its critical role in a wide range of tasks, including instance retrieval, re-identification, and identity preservation in AI-generated content. To bridge this gap, we propose a unified formulation for visual identity discrimination~(VisID) and introduce $\textbf{MVEB}$ ($\textbf{M}$ultimodal $\textbf{V}$isual Identity $\textbf{E}$mbedding $\textbf{B}$enchmark), a large-scale benchmark curated from both real-world and synthetic datasets to support evaluation and training. Furthermore, we present a simple yet effective learning framework that jointly optimizes general multimodal and visual identity representations through a carefully designed identity-aware sampling mechanism. Extensive experiments demonstrate that our approach successfully endows UMEs with strong identity discrimination capability and maintains competitive general multimodal performance. We believe this work not only illuminates a critical yet neglected capability, but also takes a step toward more holistic universal multimodal embeddings. Code and data are available at \href{https://chrisclear3.github.io/MVEB}{MVEB}.

Figures

Figures reproduced from arXiv: 2608.01794 by Bing Deng, Chaochen Gu, Jiashen Hua, Jiawei Cao, Jieping Ye, Junyi Feng, Kaijie Wu, Ziheng Huang.

Figure 1
Figure 1. Figure 1: Overview of the proposed VisME for identity-level multimodal embeddings. (a) Qualitative comparison with existing UMEs on visual identity grounding tasks; (b) Illustration of four meta-tasks in MVEB (Identity Recognition, Re-Identification, Identity Grounding, and Identity Editing), covering 28 datasets; (c) Quantitative evaluation on MVEB and MMEB benchmarks. Abstract Universal Multimodal Embeddings (UMEs… view at source ↗
Figure 2
Figure 2. Figure 2: Overall dataset curation pipeline. Identity Recognition 8.1k 21.0k 25.9k 14.0k 14.0k 25.5k 28.5k 24.0k Re- Identification 30.0k 26.4k 10.2k 17.8k 16.6k Identity Grounding 30.1k 27.4k 30.0k Identity Editing 24.0k 18.3k 18.0k 30.0k Train 439.7k Identity Recognition Cars196 (8.1k) CompCars (21.0k) Identity Recognition 1 2.4k .6k 3.5k 5.4k 2.8k 2.7k 3.0k 3.0k 3.1k 3.0k I Re- dentification 3.0k 1.8k 2.2k 1.4k 1… view at source ↗
Figure 3
Figure 3. Figure 3: The distribution of MVEB dataset. Sectors shaded in light gray correspond to out-of-domain (OOD) validation sets. [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Qualitative comparison of image retrieval results on four tasks (ID-Rec, Re-ID, ID-Grd, and ID-Edit) between our method (based [PITH_FULL_IMAGE:figures/full_fig_p008_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Performance variation with respect to batch size and [PITH_FULL_IMAGE:figures/full_fig_p008_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Our VisME model, based on the QwenVL series, freezes [PITH_FULL_IMAGE:figures/full_fig_p013_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: t-SNE of VisME-7B and VLM2Vec-7B embeddings on [PITH_FULL_IMAGE:figures/full_fig_p013_7.png] view at source ↗

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

88 extracted references · 32 canonical work pages · 1 internal anchor

  1. [1]

    Phi-3 technical report: A highly capable language model locally on your phone.arXiv preprint arXiv:2404.14219, 2024

    Marah Abdin, Jyoti Aneja, Hany Awadalla, Ahmed Awadal- lah, Ammar Ahmad Awan, Nguyen Bach, Amit Bahree, Arash Bakhtiari, Jianmin Bao, Harkirat Behl, et al. Phi-3 technical report: A highly capable language model locally on your phone.arXiv preprint arXiv:2404.14219, 2024. 2

    Pith/arXiv arXiv 2024

  2. [2]

    Unicom: Universal and compact representation learning for image re- trieval.arXiv preprint arXiv:2304.05884, 2023

    Xiang An, Jiankang Deng, Kaicheng Yang, Jaiwei Li, Ziy- ong Feng, Jia Guo, Jing Yang, and Tongliang Liu. Unicom: Universal and compact representation learning for image re- trieval.arXiv preprint arXiv:2304.05884, 2023. 3, 4

    Pith/arXiv arXiv 2023

  3. [3]

    Shuai Bai, Keqin Chen, Xuejing Liu, Jialin Wang, Wenbin Ge, Sibo Song, Kai Dang, Peng Wang, Shijie Wang, Jun Tang, et al. Qwen2. 5-vl technical report.arXiv preprint arXiv:2502.13923, 2025. 2, 3, 6

    Pith/arXiv arXiv 2025

  4. [4]

    Llm2vec: Large language models are secretly pow- erful text encoders.arXiv preprint arXiv:2404.05961, 2024

    Parishad BehnamGhader, Vaibhav Adlakha, Marius Mos- bach, Dzmitry Bahdanau, Nicolas Chapados, and Siva Reddy. Llm2vec: Large language models are secretly pow- erful text encoders.arXiv preprint arXiv:2404.05961, 2024. 2

    Pith/arXiv arXiv 2024

  5. [5]

    Flame: Frozen large language models enable data-efficient language-image pre- training

    Anjia Cao, Xing Wei, and Zhiheng Ma. Flame: Frozen large language models enable data-efficient language-image pre- training. InProceedings of the Computer Vision and Pattern Recognition Conference, pages 4080–4090, 2025. 2

    2025

  6. [6]

    mme5: Improving mul- timodal multilingual embeddings via high-quality synthetic data.arXiv preprint arXiv:2502.08468, 2025

    Haonan Chen, Liang Wang, Nan Yang, Yutao Zhu, Ziliang Zhao, Furu Wei, and Zhicheng Dou. mme5: Improving mul- timodal multilingual embeddings via high-quality synthetic data.arXiv preprint arXiv:2502.08468, 2025. 2, 3

    Pith/arXiv arXiv 2025

  7. [7]

    Murag: Multimodal retrieval-augmented generator for open question answering over images and text

    Wenhu Chen, Hexiang Hu, Xi Chen, Pat Verga, and William Cohen. Murag: Multimodal retrieval-augmented generator for open question answering over images and text. InPro- ceedings of the 2022 Conference on Empirical Methods in Natural Language Processing, pages 5558–5570, 2022. 2

    2022

  8. [8]

    Internvl: Scaling up vision foundation mod- els and aligning for generic visual-linguistic tasks

    Zhe Chen, Jiannan Wu, Wenhai Wang, Weijie Su, Guo Chen, Sen Xing, Muyan Zhong, Qinglong Zhang, Xizhou Zhu, Lewei Lu, et al. Internvl: Scaling up vision foundation mod- els and aligning for generic visual-linguistic tasks. InPro- ceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 24185–24198, 2024. 2

    2024

  9. [9]

    Opengpt-4o-image: A com- prehensive dataset for advanced image generation and edit- ing.arXiv preprint arXiv:2509.24900, 2025

    Zhihong Chen, Xuehai Bai, Yang Shi, Chaoyou Fu, Huanyu Zhang, Haotian Wang, Xiaoyan Sun, Zhang Zhang, Liang Wang, Yuanxing Zhang, et al. Opengpt-4o-image: A com- prehensive dataset for advanced image generation and edit- ing.arXiv preprint arXiv:2509.24900, 2025. 2, 6

    arXiv 2025

  10. [10]

    Think then embed: Generative context improves multimodal embedding.arXiv preprint arXiv:2510.05014,

    Xuanming Cui, Jianpeng Cheng, Hong-you Chen, Satya Narayan Shukla, Abhijeet Awasthi, Xichen Pan, Chaitanya Ahuja, Shlok Kumar Mishra, Qi Guo, Ser-Nam Lim, et al. Think then embed: Generative context improves multimodal embedding.arXiv preprint arXiv:2510.05014,

    arXiv

  11. [11]

    Arcface: Additive angular margin loss for deep face recognition

    Jiankang Deng, Jia Guo, Niannan Xue, and Stefanos Zafeiriou. Arcface: Additive angular margin loss for deep face recognition. InProceedings of the IEEE/CVF con- ference on computer vision and pattern recognition, pages 4690–4699, 2019. 2, 3

    2019

  12. [12]

    Efficient and discriminative image feature extrac- tion for universal image retrieval

    Morris Florek, David Tschirschwitz, Bj ¨orn Barz, and V olker Rodehorst. Efficient and discriminative image feature extrac- tion for universal image retrieval. InDAGM German Confer- ence on Pattern Recognition, pages 164–180. Springer, 2024. 3

    2024

  13. [13]

    Dream- sim: Learning new dimensions of human visual similar- ity using synthetic data.arXiv preprint arXiv:2306.09344,

    Stephanie Fu, Netanel Tamir, Shobhita Sundaram, Lucy Chai, Richard Zhang, Tali Dekel, and Phillip Isola. Dream- sim: Learning new dimensions of human visual similar- ity using synthetic data.arXiv preprint arXiv:2306.09344,

    Pith/arXiv arXiv

  14. [14]

    Scaling deep contrastive learning batch size under memory limited setup.arXiv preprint arXiv:2101.06983, 2021

    Luyu Gao, Yunyi Zhang, Jiawei Han, and Jamie Callan. Scaling deep contrastive learning batch size under memory limited setup.arXiv preprint arXiv:2101.06983, 2021. 6

    Pith/arXiv arXiv 2021

  15. [15]

    Seed-x: Mul- timodal models with unified multi-granularity comprehen- sion and generation.arXiv preprint arXiv:2404.14396, 2024

    Yuying Ge, Sijie Zhao, Jinguo Zhu, Yixiao Ge, Kun Yi, Lin Song, Chen Li, Xiaohan Ding, and Ying Shan. Seed-x: Mul- timodal models with unified multi-granularity comprehen- sion and generation.arXiv preprint arXiv:2404.14396, 2024. 2, 6

    Pith/arXiv arXiv 2024

  16. [16]

    Breaking the modality barrier: Universal embedding learning with multimodal llms.arXiv preprint arXiv:2504.17432, 2025

    Tiancheng Gu, Kaicheng Yang, Ziyong Feng, Xingjun Wang, Yanzhao Zhang, Dingkun Long, Yingda Chen, Wei- dong Cai, and Jiankang Deng. Breaking the modality barrier: Universal embedding learning with multimodal llms.arXiv preprint arXiv:2504.17432, 2025. 2, 3, 6

    arXiv 2025

  17. [17]

    Unime-v2: Mllm-as-a-judge for uni- versal multimodal embedding learning.arXiv preprint arXiv:2510.13515, 2025

    Tiancheng Gu, Kaicheng Yang, Kaichen Zhang, Xiang An, Ziyong Feng, Yueyi Zhang, Weidong Cai, Jiankang Deng, and Lidong Bing. Unime-v2: Mllm-as-a-judge for uni- versal multimodal embedding learning.arXiv preprint arXiv:2510.13515, 2025. 6

    arXiv 2025

  18. [18]

    jina-embeddings-v4: Universal embeddings for multimodal multilingual retrieval

    Michael G ¨unther, Saba Sturua, Mohammad Kalim Akram, Isabelle Mohr, Andrei Ungureanu, Bo Wang, Sedigheh Es- lami, Scott Martens, Maximilian Werk, Nan Wang, et al. jina-embeddings-v4: Universal embeddings for multimodal multilingual retrieval. InProceedings of the 5th Workshop on Multilingual Representation Learning (MRL 2025), pages 531–550, 2025. 2

    2025

  19. [19]

    Ms-celeb-1m: A dataset and benchmark for large-scale face recognition

    Yandong Guo, Lei Zhang, Yuxiao Hu, Xiaodong He, and Jianfeng Gao. Ms-celeb-1m: A dataset and benchmark for large-scale face recognition. InEuropean conference on computer vision, pages 87–102. Springer, 2016. 3, 2, 5

    2016

  20. [20]

    In de- fense of the triplet loss for person re-identification.arXiv preprint arXiv:1703.07737, 2017

    Alexander Hermans, Lucas Beyer, and Bastian Leibe. In de- fense of the triplet loss for person re-identification.arXiv preprint arXiv:1703.07737, 2017. 2

    Pith/arXiv arXiv 2017

  21. [21]

    Lora: Low-rank adaptation of large language models.ICLR, 1(2):3, 2022

    Edward J Hu, Yelong Shen, Phillip Wallis, Zeyuan Allen- Zhu, Yuanzhi Li, Shean Wang, Lu Wang, Weizhu Chen, et al. Lora: Low-rank adaptation of large language models.ICLR, 1(2):3, 2022. 6

    2022

  22. [22]

    Llm2clip: Powerful language model unlocks richer visual representation.arXiv preprint arXiv:2411.04997, 2024

    Weiquan Huang, Aoqi Wu, Yifan Yang, Xufang Luo, Yuqing Yang, Liang Hu, Qi Dai, Chunyu Wang, Xiyang Dai, Dongdong Chen, et al. Llm2clip: Powerful language model unlocks richer visual representation.arXiv preprint arXiv:2411.04997, 2024. 2, 3, 6

    arXiv 2024

  23. [23]

    Scaling up visual and vision-language representa- tion learning with noisy text supervision

    Chao Jia, Yinfei Yang, Ye Xia, Yi-Ting Chen, Zarana Parekh, Hieu Pham, Quoc Le, Yun-Hsuan Sung, Zhen Li, and Tom Duerig. Scaling up visual and vision-language representa- tion learning with noisy text supervision. InInternational conference on machine learning, pages 4904–4916. PMLR,

  24. [24]

    E5-v: Universal embeddings with multimodal large language models.arXiv preprint arXiv:2407.12580, 2024

    Ting Jiang, Minghui Song, Zihan Zhang, Haizhen Huang, Weiwei Deng, Feng Sun, Qi Zhang, Deqing Wang, and Fuzhen Zhuang. E5-v: Universal embeddings with multimodal large language models.arXiv preprint arXiv:2407.12580, 2024. 3

    Pith/arXiv arXiv 2024

  25. [25]

    Vlm2vec: Training vision-language models for massive multimodal embedding tasks

    Ziyan Jiang, Rui Meng, Xinyi Yang, Semih Yavuz, Yingbo Zhou, and Wenhu Chen. Vlm2vec: Training vision-language models for massive multimodal embedding tasks. InThe Thirteenth International Conference on Learning Represen- tations. 2, 3, 6

    work page

  26. [26]

    Modality curation: Building uni- versal embeddings for advanced multimodal information re- trieval.arXiv preprint arXiv:2505.19650, 2025

    Fanheng Kong, Jingyuan Zhang, Yahui Liu, Hongzhi Zhang, Shi Feng, Xiaocui Yang, Daling Wang, Yu Tian, Fuzheng Zhang, Guorui Zhou, et al. Modality curation: Building uni- versal embeddings for advanced multimodal information re- trieval.arXiv preprint arXiv:2505.19650, 2025. 2, 3

    Pith/arXiv arXiv 2025

  27. [27]

    Ilias: Instance-level image retrieval at scale

    Giorgos Kordopatis-Zilos, Vladan Stojni ´c, Anna Manko, Pavel Suma, Nikolaos-Antonios Ypsilantis, Nikos Efthymi- adis, Zakaria Laskar, Jiri Matas, Ondrej Chum, and Gior- gos Tolias. Ilias: Instance-level image retrieval at scale. In Proceedings of the Computer Vision and Pattern Recognition Conference, pages 14777–14787, 2025. 2, 3

    work page 2025

  28. [28]

    jina- clip-v2: Multilingual multimodal embeddings for text and images.arXiv preprint arXiv:2412.08802, 2024

    Andreas Koukounas, Georgios Mastrapas, Sedigheh Eslami, Bo Wang, Mohammad Kalim Akram, Michael G ¨unther, Is- abelle Mohr, Saba Sturua, Nan Wang, and Han Xiao. jina- clip-v2: Multilingual multimodal embeddings for text and images.arXiv preprint arXiv:2412.08802, 2024. 2

    Pith/arXiv arXiv 2024

  29. [29]

    3d object representations for fine-grained categorization

    Jonathan Krause, Michael Stark, Jia Deng, and Li Fei-Fei. 3d object representations for fine-grained categorization. In Proceedings of the IEEE international conference on com- puter vision workshops, pages 554–561, 2013. 3, 2, 4

    work page 2013

  30. [30]

    Llave: Large language and vision embedding models with hardness-weighted contrastive learning.arXiv preprint arXiv:2503.04812, 2025

    Zhibin Lan, Liqiang Niu, Fandong Meng, Jie Zhou, and Jinsong Su. Llave: Large language and vision embedding models with hardness-weighted contrastive learning.arXiv preprint arXiv:2503.04812, 2025. 3, 6

    arXiv 2025

  31. [31]

    Nv- embed: Improved techniques for training llms as generalist embedding models.arXiv preprint arXiv:2405.17428, 2024

    Chankyu Lee, Rajarshi Roy, Mengyao Xu, Jonathan Raiman, Mohammad Shoeybi, Bryan Catanzaro, and Wei Ping. Nv- embed: Improved techniques for training llms as generalist embedding models.arXiv preprint arXiv:2405.17428, 2024. 2

    Pith/arXiv arXiv 2024

  32. [32]

    Llava-onevision: Easy visual task transfer

    Bo Li, Yuanhan Zhang, Dong Guo, Renrui Zhang, Feng Li, Hao Zhang, Kaichen Zhang, Peiyuan Zhang, Yanwei Li, Zi- wei Liu, et al. Llava-onevision: Easy visual task transfer. arXiv preprint arXiv:2408.03326, 2024. 2, 3

    Pith/arXiv arXiv 2024

  33. [33]

    MagicID: Hybrid Preference Optimization for ID-Consistent and Dynamic-Preserved Video Customization

    Hengjia Li, Lifan Jiang, Xi Xiao, Tianyang Wang, Hongwei Yi, Boxi Wu, and Deng Cai. Magicid: Hybrid preference optimization for id-consistent and dynamic-preserved video customization.arXiv preprint arXiv:2503.12689, 2025. 2, 3

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  34. [34]

    Personalvideo: High id-fidelity video customization without dynamic and semantic degradation

    Hengjia Li, Haonan Qiu, Shiwei Zhang, Xiang Wang, Yu- jie Wei, Zekun Li, Yingya Zhang, Boxi Wu, and Deng Cai. Personalvideo: High id-fidelity video customization without dynamic and semantic degradation. InProceedings of the IEEE/CVF International Conference on Computer Vision, pages 19406–19416, 2025. 2, 3

    work page 2025

  35. [35]

    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. InInterna- tional conference on machine learning, pages 12888–12900. PMLR, 2022. 2

    2022

  36. [36]

    Open-sora plan: Open-source large video generation model.arXiv preprint arXiv:2412.00131, 2024

    Bin Lin, Yunyang Ge, Xinhua Cheng, Zongjian Li, Bin Zhu, Shaodong Wang, Xianyi He, Yang Ye, Shenghai Yuan, Li- uhan Chen, et al. Open-sora plan: Open-source large video generation model.arXiv preprint arXiv:2412.00131, 2024. 3, 4

    Pith/arXiv arXiv 2024

  37. [37]

    Uniworld: High-resolution semantic en- coders for unified visual understanding and generation.arXiv preprint arXiv:2506.03147, 2025

    Bin Lin, Zongjian Li, Xinhua Cheng, Yuwei Niu, Yang Ye, Xianyi He, Shenghai Yuan, Wangbo Yu, Shaodong Wang, Yunyang Ge, et al. Uniworld: High-resolution semantic en- coders for unified visual understanding and generation.arXiv preprint arXiv:2506.03147, 2025. 2, 6

    Pith/arXiv arXiv 2025

  38. [38]

    Mm-embed: Universal multimodal retrieval with multimodal llms

    Sheng-Chieh Lin, Chankyu Lee, Mohammad Shoeybi, Jimmy Lin, Bryan Catanzaro, and Wei Ping. Mm-embed: Universal multimodal retrieval with multimodal llms. InThe Thirteenth International Conference on Learning Represen- tations. 2, 3

    work page

  39. [39]

    Idmr: Towards instance-driven precise visual cor- respondence in multimodal retrieval.arXiv preprint arXiv:2504.00954, 2025

    Bangwei Liu, Yicheng Bao, Shaohui Lin, Xuhong Wang, Xin Tan, Yingchun Wang, Yuan Xie, and Chaochao Lu. Idmr: Towards instance-driven precise visual cor- respondence in multimodal retrieval.arXiv preprint arXiv:2504.00954, 2025. 3, 2, 6

    Pith/arXiv arXiv 2025

  40. [40]

    Automatic synthetic data and fine- grained adaptive feature alignment for composed person re- trieval

    Delong Liu, Haiwen Li, Zhaohui Hou, Zhicheng Zhao, Fei Su, and Yuan Dong. Automatic synthetic data and fine- grained adaptive feature alignment for composed person re- trieval. InThe Thirty-ninth Annual Conference on Neural Information Processing Systems, 2025. 2, 3, 6

    work page 2025

  41. [41]

    Sphereface: Deep hypersphere embedding for face recognition

    Weiyang Liu, Yandong Wen, Zhiding Yu, Ming Li, Bhiksha Raj, and Le Song. Sphereface: Deep hypersphere embedding for face recognition. InProceedings of the IEEE conference on computer vision and pattern recognition, pages 212–220,

  42. [42]

    Large- scale vehicle re-identification in urban surveillance videos

    Xinchen Liu, Wu Liu, Huadong Ma, and Huiyuan Fu. Large- scale vehicle re-identification in urban surveillance videos. In2016 IEEE international conference on multimedia and expo (ICME), pages 1–6. IEEE, 2016. 2, 5

    work page 2016

  43. [43]

    Lamra: Large multimodal model as your advanced retrieval assistant

    Yikun Liu, Yajie Zhang, Jiayin Cai, Xiaolong Jiang, Yao Hu, Jiangchao Yao, Yanfeng Wang, and Weidi Xie. Lamra: Large multimodal model as your advanced retrieval assistant. In Proceedings of the Computer Vision and Pattern Recognition Conference, pages 4015–4025, 2025. 2, 6

    work page 2025

  44. [44]

    Deepfashion: Powering robust clothes recognition and retrieval with rich annotations

    Ziwei Liu, Ping Luo, Shi Qiu, Xiaogang Wang, and Xiaoou Tang. Deepfashion: Powering robust clothes recognition and retrieval with rich annotations. InProceedings of the IEEE conference on computer vision and pattern recogni- tion, pages 1096–1104, 2016. 2, 4

    work page 2016

  45. [45]

    Vlm2vec-v2: Advancing multimodal em- bedding for videos, images, and visual documents.arXiv preprint arXiv:2507.04590, 2025

    Rui Meng, Ziyan Jiang, Ye Liu, Mingyi Su, Xinyi Yang, Yuepeng Fu, Can Qin, Zeyuan Chen, Ran Xu, Caiming Xiong, et al. Vlm2vec-v2: Advancing multimodal em- bedding for videos, images, and visual documents.arXiv preprint arXiv:2507.04590, 2025. 2, 3

    Pith/arXiv arXiv 2025

  46. [46]

    A culturally-aware benchmark for person re-identification in modest attire.Engineering Ap- plications of Artificial Intelligence, 158:111494, 2025

    Alireza Sedighi Moghaddam, Fatemeh Anvari, Mohammad- javad Mirshekari Haghighi, Mohammadali Fakhari, and Mo- hammad Reza Mohammadi. A culturally-aware benchmark for person re-identification in modest attire.Engineering Ap- plications of Artificial Intelligence, 158:111494, 2025. 2, 5

    work page 2025

  47. [48]

    Deep metric learning via lifted structured fea- ture embedding

    Hyun Oh Song, Yu Xiang, Stefanie Jegelka, and Silvio Savarese. Deep metric learning via lifted structured fea- ture embedding. InProceedings of the IEEE conference on computer vision and pattern recognition, pages 4004–4012,

    work page

  48. [49]

    Rp2k: A large- scale retail product dataset for fine-grained image classifica- tion.arXiv preprint arXiv:2006.12634, 2020

    Jingtian Peng, Chang Xiao, and Yifan Li. Rp2k: A large- scale retail product dataset for fine-grained image classifica- tion.arXiv preprint arXiv:2006.12634, 2020. 3, 2, 4

    Pith/arXiv arXiv 2006

  49. [50]

    Revisiting oxford and paris: Large-scale image retrieval benchmarking

    Filip Radenovi ´c, Ahmet Iscen, Giorgos Tolias, Yannis Avrithis, and Ond ˇrej Chum. Revisiting oxford and paris: Large-scale image retrieval benchmarking. InProceedings of the IEEE conference on computer vision and pattern recog- nition, pages 5706–5715, 2018. 3

    work page 2018

  50. [51]

    Learning transferable visual models from natural language supervi- sion

    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 supervi- sion. InInternational conference on machine learning, pages 8748–8763. PmLR, 2021. 2, 3

    2021

  51. [52]

    Performance measures and a data set for multi-target, multi-camera tracking

    Ergys Ristani, Francesco Solera, Roger Zou, Rita Cucchiara, and Carlo Tomasi. Performance measures and a data set for multi-target, multi-camera tracking. InEuropean conference on computer vision, pages 17–35. Springer, 2016. 3, 2, 5

    work page 2016

  52. [53]

    Eva-clip: Improved training techniques for clip at scale

    Quan Sun, Yuxin Fang, Ledell Wu, Xinlong Wang, and Yue Cao. Eva-clip: Improved training techniques for clip at scale. arXiv preprint arXiv:2303.15389, 2023. 2

    Pith/arXiv arXiv 2023

  53. [54]

    Visual named entity linking: A new dataset and a baseline

    Wen Sun, Yixing Fan, Jiafeng Guo, Ruqing Zhang, and Xueqi Cheng. Visual named entity linking: A new dataset and a baseline. InFindings of the Association for Compu- tational Linguistics: EMNLP 2022, pages 2403–2415, 2022. 2, 6

    work page 2022

  54. [55]

    Breaking the batch barrier (b3) of contrastive learning via smart batch mining.arXiv preprint arXiv:2505.11293, 2025

    Raghuveer Thirukovalluru, Rui Meng, Ye Liu, Mingyi Su, Ping Nie, Semih Yavuz, Yingbo Zhou, Wenhu Chen, Bhuwan Dhingra, et al. Breaking the batch barrier (b3) of contrastive learning via smart batch mining.arXiv preprint arXiv:2505.11293, 2025. 2, 3, 6

    arXiv 2025

  55. [56]

    Siglip 2: Multilingual vision-language en- coders with improved semantic understanding, localization, and dense features.arXiv preprint arXiv:2502.14786, 2025

    Michael Tschannen, Alexey Gritsenko, Xiao Wang, Muham- mad Ferjad Naeem, Ibrahim Alabdulmohsin, Nikhil Parthasarathy, Talfan Evans, Lucas Beyer, Ye Xia, Basil Mustafa, et al. Siglip 2: Multilingual vision-language en- coders with improved semantic understanding, localization, and dense features.arXiv preprint arXiv:2502.14786, 2025. 2, 6

    Pith/arXiv arXiv 2025

  56. [57]

    The inaturalist species classification and de- tection dataset

    Grant Van Horn, Oisin Mac Aodha, Yang Song, Yin Cui, Chen Sun, Alex Shepard, Hartwig Adam, Pietro Perona, and Serge Belongie. The inaturalist species classification and de- tection dataset. InProceedings of the IEEE conference on computer vision and pattern recognition, pages 8769–8778,

  57. [58]

    Qwen2-vl: Enhancing vision-language model’s perception of the world at any resolution.arXiv preprint arXiv:2409.12191, 2024

    Peng Wang, Shuai Bai, Sinan Tan, Shijie Wang, Zhihao Fan, Jinze Bai, Keqin Chen, Xuejing Liu, Jialin Wang, Wenbin Ge, et al. Qwen2-vl: Enhancing vision-language model’s perception of the world at any resolution.arXiv preprint arXiv:2409.12191, 2024. 2, 3, 6

    Pith/arXiv arXiv 2024

  58. [59]

    Instantid: Zero-shot identity-preserving generation in seconds.arXiv preprint arXiv:2401.07519, 2024

    Qixun Wang, Xu Bai, Haofan Wang, Zekui Qin, Anthony Chen, Huaxia Li, Xu Tang, and Yao Hu. Instantid: Zero-shot identity-preserving generation in seconds.arXiv preprint arXiv:2401.07519, 2024. 2, 3

    Pith/arXiv arXiv 2024

  59. [60]

    Gpt-image-edit- 1.5 m: A million-scale, gpt-generated image dataset.arXiv preprint arXiv:2507.21033, 2025

    Yuhan Wang, Siwei Yang, Bingchen Zhao, Letian Zhang, Qing Liu, Yuyin Zhou, and Cihang Xie. Gpt-image-edit- 1.5 m: A million-scale, gpt-generated image dataset.arXiv preprint arXiv:2507.21033, 2025. 4, 2, 6

    Pith/arXiv arXiv 2025

  60. [61]

    Uniir: Train- ing and benchmarking universal multimodal information re- trievers

    Cong Wei, Yang Chen, Haonan Chen, Hexiang Hu, Ge Zhang, Jie Fu, Alan Ritter, and Wenhu Chen. Uniir: Train- ing and benchmarking universal multimodal information re- trievers. InEuropean Conference on Computer Vision, pages 387–404. Springer, 2024. 2, 3

    work page 2024

  61. [62]

    Google landmarks dataset v2-a large-scale benchmark for instance-level recognition and retrieval

    Tobias Weyand, Andre Araujo, Bingyi Cao, and Jack Sim. Google landmarks dataset v2-a large-scale benchmark for instance-level recognition and retrieval. InProceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 2575–2584, 2020. 3, 4, 2

    work page 2020

  62. [63]

    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. InProceedings of the IEEE/CVF Conference on computer vision and pattern recognition, pages 11307– 11317, 2021. 2

    work page 2021

  63. [64]

    Forb: a flat object retrieval benchmark for universal im- age embedding.Advances in Neural Information Processing Systems, 36:25448–25460, 2023

    Pengxiang Wu, Siman Wang, Kevin Dela Rosa, and Derek Hu. Forb: a flat object retrieval benchmark for universal im- age embedding.Advances in Neural Information Processing Systems, 36:25448–25460, 2023. 2, 6

    work page 2023

  64. [65]

    Withanyone: Towards control- lable and id consistent image generation.arXiv preprint arXiv:2510.14975, 2025

    Hengyuan Xu, Wei Cheng, Peng Xing, Yixiao Fang, Shuhan Wu, Rui Wang, Xianfang Zeng, Daxin Jiang, Gang Yu, Xingjun Ma, et al. Withanyone: Towards control- lable and id consistent image generation.arXiv preprint arXiv:2510.14975, 2025. 2, 6

    arXiv 2025

  65. [66]

    Improve multi-modal embedding learning via explicit hard negative gradient am- plifying.arXiv preprint arXiv:2506.02020, 2025

    Youze Xue, Dian Li, and Gang Liu. Improve multi-modal embedding learning via explicit hard negative gradient am- plifying.arXiv preprint arXiv:2506.02020, 2025. 2, 3

    Pith/arXiv arXiv 2025

  66. [67]

    Chinese clip: Contrastive vision-language pretraining in chinese.arXiv preprint arXiv:2211.01335, 2022

    An Yang, Junshu Pan, Junyang Lin, Rui Men, Yichang Zhang, Jingren Zhou, and Chang Zhou. Chinese clip: Contrastive vision-language pretraining in chinese.arXiv preprint arXiv:2211.01335, 2022. 2

    Pith/arXiv arXiv 2022

  67. [68]

    A large-scale car dataset for fine-grained categorization and verification

    Linjie Yang, Ping Luo, Chen Change Loy, and Xiaoou Tang. A large-scale car dataset for fine-grained categorization and verification. InProceedings of the IEEE conference on computer vision and pattern recognition, pages 3973–3981,

    work page

  68. [69]

    Deep learning for person re- identification: A survey and outlook.IEEE transactions on pattern analysis and machine intelligence, 44(6):2872–2893,

    Mang Ye, Jianbing Shen, Gaojie Lin, Tao Xiang, Ling Shao, and Steven CH Hoi. Deep learning for person re- identification: A survey and outlook.IEEE transactions on pattern analysis and machine intelligence, 44(6):2872–2893,

    work page

  69. [70]

    Learn- ing face representation from scratch.arXiv preprint arXiv:1411.7923, 2014

    Dong Yi, Zhen Lei, Shengcai Liao, and Stan Z Li. Learn- ing face representation from scratch.arXiv preprint arXiv:1411.7923, 2014. 2, 5

    Pith/arXiv arXiv 2014

  70. [71]

    The met dataset: Instance-level recognition for artworks

    Nikolaos-Antonios Ypsilantis, Noa Garcia, Guangxing Han, Sarah Ibrahimi, Nanne Van Noord, and Giorgos Tolias. The met dataset: Instance-level recognition for artworks. In Thirty-fifth conference on neural information processing sys- tems datasets and benchmarks track (Round 2), 2021. 2, 4

    work page 2021

  71. [72]

    Towards universal image embed- dings: A large-scale dataset and challenge for generic image representations

    Nikolaos-Antonios Ypsilantis, Kaifeng Chen, Bingyi Cao, M´ario Lipovsk `y, Pelin Dogan-Sch ¨onberger, Grzegorz Makosa, Boris Bluntschli, Mojtaba Seyedhosseini, Ond ˇrej Chum, and Andr´e Araujo. Towards universal image embed- dings: A large-scale dataset and challenge for generic image representations. InProceedings of the ieee/cvf international conference...

    work page 2023

  72. [73]

    Udon: Universal dynamic online distilla- tion for generic image representations.Advances in Neural Information Processing Systems, 37:86836–86859, 2024

    Nikolaos-Antonios Ypsilantis, Kaifeng Chen, Andr ´e Araujo, and Ondˇrej Chum. Udon: Universal dynamic online distilla- tion for generic image representations.Advances in Neural Information Processing Systems, 37:86836–86859, 2024. 2, 3, 4

    work page 2024

  73. [74]

    Coca: Contrastive captioners are image-text foundation models.arXiv preprint arXiv:2205.01917, 2022

    Jiahui Yu, Zirui Wang, Vijay Vasudevan, Legg Yeung, Mo- jtaba Seyedhosseini, and Yonghui Wu. Coca: Contrastive captioners are image-text foundation models.arXiv preprint arXiv:2205.01917, 2022. 2

    Pith/arXiv arXiv 2022

  74. [75]

    Visrag: Vision-based retrieval-augmented generation on multi-modality documents.arXiv preprint arXiv:2410.10594, 2024

    Shi Yu, Chaoyue Tang, Bokai Xu, Junbo Cui, Junhao Ran, Yukun Yan, Zhenghao Liu, Shuo Wang, Xu Han, Zhiyuan Liu, et al. Visrag: Vision-based retrieval-augmented generation on multi-modality documents.arXiv preprint arXiv:2410.10594, 2024. 2

    Pith/arXiv arXiv 2024

  75. [76]

    Identity- preserving text-to-video generation by frequency decompo- sition

    Shenghai Yuan, Jinfa Huang, Xianyi He, Yunyang Ge, Yu- jun Shi, Liuhan Chen, Jiebo Luo, and Li Yuan. Identity- preserving text-to-video generation by frequency decompo- sition. InProceedings of the Computer Vision and Pattern Recognition Conference, pages 12978–12988, 2025. 2, 3

    work page 2025

  76. [77]

    Sigmoid loss for language image pre-training

    Xiaohua Zhai, Basil Mustafa, Alexander Kolesnikov, and Lucas Beyer. Sigmoid loss for language image pre-training. InProceedings of the IEEE/CVF international conference on computer vision, pages 11975–11986, 2023. 2

    2023

  77. [78]

    Prod- uct1m: Towards weakly supervised instance-level product retrieval via cross-modal pretraining

    Xunlin Zhan, Yangxin Wu, Xiao Dong, Yunchao Wei, Min- long Lu, Yichi Zhang, Hang Xu, and Xiaodan Liang. Prod- uct1m: Towards weakly supervised instance-level product retrieval via cross-modal pretraining. InProceedings of the IEEE/CVF international conference on computer vision, pages 11782–11791, 2021. 2, 4

    work page 2021

  78. [79]

    Magi- clens: Self-supervised image retrieval with open-ended in- structions.arXiv preprint arXiv:2403.19651, 2024

    Kai Zhang, Yi Luan, Hexiang Hu, Kenton Lee, Siyuan Qiao, Wenhu Chen, Yu Su, and Ming-Wei Chang. Magi- clens: Self-supervised image retrieval with open-ended in- structions.arXiv preprint arXiv:2403.19651, 2024. 2

    Pith/arXiv arXiv 2024

  79. [80]

    Assess- ing and learning alignment of unimodal vision and language models

    Le Zhang, Qian Yang, and Aishwarya Agrawal. Assess- ing and learning alignment of unimodal vision and language models. InProceedings of the Computer Vision and Pattern Recognition Conference, pages 14604–14614, 2025. 2

    2025

  80. [81]

    Beyond frontal faces: Improving per- son recognition using multiple cues

    Ning Zhang, Manohar Paluri, Yaniv Taigman, Rob Fergus, and Lubomir Bourdev. Beyond frontal faces: Improving per- son recognition using multiple cues. InProceedings of the IEEE conference on computer vision and pattern recogni- tion, pages 4804–4813, 2015. 2, 6

    work page 2015

Showing first 80 references.

This paper was first reviewed by deepseek-v4-flash on August 4, 2026.