Pith. sign in

REVIEW 4 major objections 5 minor 56 references

Enhancing Visual Representation for Text-based Person Searching

T0 review · 4 major / 5 minor · reviewed 2026-08-10 · deepseek-v4-flash

Pith's one-line read CLIP knowledge transfers to person search with two training-only tasks, yielding top results on three benchmarks.

desk verdict Useful training-only auxiliary losses with honest ablations, but the SOTA claim is contradicted by their own RaSa comparison and gains lack variance estimates. read the letter →

arxiv 2412.20646 v1 pith:MST32T3G submitted 2024-12-30 cs.CV

classification cs.CV
keywords text-basedpersonsearchmaskedimagemodelingCLIPfine-tuningidentity-awarevisualfeaturescross-modalretrievalre-identificationauxiliarytrainingtasksvision-languagemodel
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

Text-based person search — retrieving the pedestrian in a large image database that matches a natural-language description — has been limited by models' weak grasp of visual details and by identity confusion, not by lack of cross-modal alignment machinery. This paper claims that two training-only auxiliary tasks, applied on top of a fine-tuned CLIP backbone, solve both problems: a text-guided masked image modeling task forces the image encoder to reconstruct masked patches using textual cues, and an identity-supervised distribution-matching task pulls same-identity global features together while pushing different-identity features apart. With these tasks, the model needs only global feature similarity at inference, yet reaches Rank-1 accuracy of 72.47%, 62.71%, and 59.25% on CUHK-PEDES, ICFG-PEDES, and RSTPReid, surpassing every model in its comparison tables. The paper's key finding is that CLIP's pretrained multimodal knowledge can be successfully adapted to pedestrian search when guided by these auxiliary tasks, without complex local alignment or segmentation priors.

What carries the argument

The two load-bearing mechanisms are auxiliary losses active only during training. Text Guided Masked Image Modeling (TG-MIM) is a masked-image-modeling objective in which randomly masked image patches are reconstructed from a representation produced by multi-head cross-modal attention, where the text query supplies keys and values; the $L_1$ pixel-reconstruction loss forces the image encoder to retain local visual details that are useful for the query. Identity Supervised Global Visual Feature Calibration (IS-GVFC) is a KL-divergence loss between the soft image-to-image matching probabilities predicted from global visual feature similarities and the ground-truth probability distribution over identities in the batch, which compresses same-identity feature clusters and separates different-identity ones. Together they adapt a CLIP-initialized backbone to the pedestrian domain, and because both branches are removed at inference, the deployed model is just a CLIP encoder pair with a cosine-similarity head.

What would settle it

Train VFE-TPS with the TG-MIM loss removed but all other components unchanged, repeat over multiple random seeds, and check whether the Rank-1 gap on CUHK-PEDES is reproducibly larger than about one percentage point; if the gap disappears or flips sign, the central mechanism is unsupported.

Watch

Extended reading notes

Core claim

The paper's central claim is that visual feature quality, not the complexity of alignment, is the bottleneck in text-based person search. VFE-TPS initializes image and text encoders from CLIP, then adds two training-only objectives. TG-MIM randomly masks image patches, lets a cross-modal attention layer condition visual features on the text query, and reconstructs the masked raw pixels under an L1 loss, teaching the image encoder to capture detail that matters for the query. IS-GVFC samples image pairs in a batch, computes soft matching probabilities from global visual feature similarities, and minimizes the KL divergence to the ground-truth identity matching distribution, making global features identity-aware. During inference both auxiliary branches are discarded, so retrieval reduces to cosine similarity between global image and text features. The paper reports that this recipe lifts its CLIP baseline from 70.61% to 72.47% Rank-1 on CUHK-PEDES, and beats all methods in its main comparison on all three benchmarks.

Load-bearing premise

The claim depends on the premise that teaching the model to reconstruct randomly masked image pixels under text guidance transfers to the ability to spot the fine-grained visual details that distinguish one pedestrian identity from another, rather than just low-level texture and color statistics.

Editorial extensions

If this is right

  • Person search models can discard complex local-alignment modules and still improve accuracy, because global features already contain the task-relevant local information after auxiliary training.
  • CLIP's multimodal knowledge, though learned on natural images, is adaptable to pedestrian retrieval through training-time auxiliary objectives rather than architectural changes.
  • The two auxiliary tasks are complementary: TG-MIM improves detail understanding, IS-GVFC improves identity awareness, and their joint use gives the best results (72.47% Rank-1 vs 70.61% baseline on CUHK-PEDES).
  • Inference becomes simpler and cheaper than local-alignment competitors, since only global features are extracted and compared.

Reading between the lines

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

  • A natural testable extension is to apply TG-MIM and IS-GVFC to other attribute-based instance retrieval tasks, such as vehicle search or product retrieval; the paper does not report such experiments.
  • Because TG-MIM is ablated only on point estimates, a multi-seed paired comparison would clarify how much of the Rank-1 gain is attributable to pixel reconstruction versus other training details; this is an inference, not a claim in the paper.
  • The recognized failure on vague queries could be addressed by an interactive system that lets users refine the description based on top-ranked candidates, a direction the paper mentions but does not implement.
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

4 major / 5 minor

Summary. The manuscript proposes VFE-TPS, a text-based person search model built on CLIP image/text encoders, with two training-only auxiliary tasks: Text Guided Masked Image Modeling (TG-MIM), which reconstructs masked image patches using cross-attention with the text query, and Identity Supervised Global Visual Feature Calibration (IS-GVFC), which uses a KL divergence between visual feature similarity distributions and identity-label distributions. The authors report Rank-1 accuracy of 72.47% on CUHK-PEDES, 62.71% on ICFG-PEDES, and 59.25% on RSTPReid, and argue that the auxiliary tasks enhance the image encoder's local detail understanding and identity awareness, enabling successful adaptation of CLIP to the text-based person search task. The paper includes comparisons with prior methods, ablations, computational complexity analysis, and qualitative retrieval visualizations.

Significance. If the attribution claims are established, the paper would make a useful contribution by showing that CLIP can be adapted to text-based person search through lightweight, annotation-free auxiliary tasks, avoiding complex local alignment during inference. The authors ship code, evaluate on three standard benchmarks, and include multiple ablations, which are strengths. However, the headline claim of exceeding existing approaches is not supported by the paper's own Table 4, where RaSa reports Rank-1 76.51% versus the proposed 72.47% on CUHK-PEDES, and the central mechanism claim rests on single-run point estimates without variance. The proposed method is plausible and the ideas are worth publishing once the scope claims and the attribution evidence are tightened.

major comments (4)
  1. [Section 4.4, Table 4] The claim that the model 'surpassing all the SOTAs' (Section 4.4) and the abstract's 'exceeds the existing approaches' are contradicted by Table 4, which lists RaSa with Rank-1 76.51% on CUHK-PEDES, higher than VFE-TPS's 72.47%. The text acknowledges RaSa but only discusses efficiency; the superiority claim must be explicitly scoped to models with the same backbone or to an accuracy-efficiency trade-off. In addition, the text cites specific GFLOPs values for RaSa and VFE-TPS, but Table 4 does not report GFLOPs; either add these numbers with a measurement protocol/source or remove them.
  2. [Sections 4.5.1 and 4.5.2, Tables 5 and 6] The baseline used to attribute the gains is inconsistent. Table 5 reports 'CLIP' fine-tuned with CMPM loss at Rank-1 66.78%, while Table 6's no-auxiliary baseline is 70.61%, and Table 1's 'baseline (ours)' is described as ViT+BERT with Rank-1 70.12%. If these are different training setups, the differences must be stated explicitly; as written, the improvement attributed to the auxiliary tasks is ambiguous (1.86 points from the 70.61 baseline, not 5.69 points implied by the gap between Table 5's CLIP row and the final model). Furthermore, all results are single-run point estimates with no variance, so differences of 0.55-1.86 Rank-1 points are within the range of typical random-seed variation for this task; the authors should report multiple seeds or significance tests.
  3. [Section 3.2.1, Table 6, Figure 5] The causal mechanism of TG-MIM is not isolated by the provided ablations. Comparing TG-MIM against MAE and SimMIM changes both the prediction target and the use of the text-guided cross-attention module, so the comparison does not identify whether the raw-pixel reconstruction signal or the added module/capacity produces the gains. A control that keeps the same TG-MIM architecture but uses a constant or non-informative reconstruction target would separate the reconstruction signal from extra regularization/capacity. Also, the implementation details (Section 4.3) do not state the final masking ratio or the temperature tau used in Equations (5) and (6); Figure 6 indicates a range of 0.5-0.6 for the masking ratio but the deployed value is not given, which is necessary for reproducibility of the central results.
  4. [Section 4.5.4, Figure 7] The comparison of IS-GVFC with ID loss and triplet loss is presented only through silhouette coefficient and t-SNE visualizations; no Rank-1/mAP results are reported for the same training budget with ID loss or triplet loss. Since the paper claims that IS-GVFC is superior to these common calibration losses, retrieval metrics for those alternatives are needed. Moreover, Table 6 shows that Rank-5 with both auxiliary tasks (88.24%) is lower than with TG-MIM alone (88.47%), so the joint effect of the two tasks is less clean than the text suggests and should be discussed or re-run.
minor comments (5)
  1. [Throughout] Typographical errors should be corrected: 'propopsed' (Section 3.3), 'promissing' (Section 4.5.1), 'rather then' (Section 1), 'In the recent decay' (Section 1), and 'KL distance' used for KL divergence in Algorithm 2.
  2. [Figure 8 caption] The panels in the Figure 8 caption are labeled '(3)' and '(4)' instead of '(c)' and '(d)', and the sentence after the figure should be updated to match the panel labels.
  3. [Algorithm 1 and Equation (3)] The loss symbol in Algorithm 1 is written as L_TD-MIM, while Equation (3) defines L_TG-MIM; the notation should be made consistent.
  4. [Figures 5 and 6] The improvement of TG-MIM over MAE and SimMIM is reported as point curves without error bars or the number of repetitions; the statement that TG-MIM is 'significantly lower' in avgDist is not supported by the plot alone.
  5. [References [31] and [37]] References [31] and [37] appear to refer to the same paper (CLIP-driven fine-grained text-image person re-identification) with different years and venues; the citation list should be deduplicated and verified.

Circularity Check

0 steps flagged · score 2.0 of 10

No significant circularity: retrieval gains are externally benchmarked; the sole self-citation is not load-bearing.

full rationale

The paper's main claims are validated on three public benchmarks (CUHK-PEDES, ICFG-PEDES, RSTPReid) with Rank-1 numbers that are not fitted parameters or renamed training losses. The two auxiliary tasks, TG-MIM and IS-GVFC, are ablated on the same retrieval metrics they are meant to improve; this is empirical validation rather than circular derivation, because the reported accuracies come from training on public data and are not equal to the auxiliary losses by construction. The TG-MIM pixel-reconstruction objective (Eq. 3) is a training proxy, and the paper does not equate reconstruction error with retrieval accuracy mathematically. The only author-overlapping citation is [10], used in Section 3.2.2 to support the claim that ID loss and triplet loss inadequately calibrate visual features; however, Section 4.5.4 independently compares ID loss, triplet loss, and IS-GVFC using silhouette coefficient and t-SNE, so the self-citation is not load-bearing. The inconsistency between Table 5's CLIP fine-tuning baseline (66.78) and Table 6's no-auxiliary baseline (70.61) is an experimental-reporting issue that may affect the attribution of gains, but it is not a circular argument. No equation is shown to reduce to its own input, and no fitted value is renamed as a prediction.

Assumptions & free parameters 3 free parameters · 4 assumptions · 0 invented entities

The central empirical results rest on two auxiliary objectives, neither of which is derived from first principles. The main unproved premises are that text-guided pixel reconstruction improves retrieval and that KL-based identity calibration is better than ID/triplet losses. No new physical entities are introduced.

free parameters (3)
  • temperature tau (Eq. 5 and Eq. 6) = not reported
    Controls the sharpness of predicted matching probabilities in IS-GVFC and CMPM. Chosen by hand, and its numerical value is not stated in the paper.
  • image masking ratio = around 0.5 to 0.6
    Selected after ablation on CUHK-PEDES in Figure 6(a). Tuning this on the same benchmark whose headline number is reported weakens the claim that TG-MIM alone drives the gain.
  • sampled image pair count C = 20
    Number of image pairs used in each IS-GVFC KL computation. A design choice that affects the loss estimate and training variance.
assumptions (4)
  • domain assumption CLIP pre-trained weights contain transferable cross-modal knowledge for pedestrian retrieval.
    Invoked in Section 3.1 and analyzed in Section 4.5.1. If this were false, fine-tuning from CLIP would not provide a benefit over unimodal encoders.
  • ad hoc to paper Raw pixel reconstruction loss (L1 on masked patches) improves the image encoder's ability to extract task-relevant visual details.
    Core assumption behind TG-MIM in Section 3.2.1. Not proven theoretically, only supported by ablation point estimates.
  • ad hoc to paper KL divergence between visual feature similarity distribution and identity-label distribution is a valid surrogate for identity-aware feature calibration.
    Underpins IS-GVFC in Section 3.2.2. The paper compares it empirically to ID loss and triplet loss but gives no formal justification.
  • domain assumption Identity labels in the training data are reliable and sufficient for supervising visual feature calibration.
    The IS-GVFC loss uses person identity labels from CUHK-PEDES, ICFG-PEDES, and RSTPReid. If labels are noisy, the calibration signal is degraded.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Enhancing Visual Representation for Text-based Person Searching." pith.science (2026). https://pith.science/paper/MST32T3G

@misc{pith2026241220646,
  author       = {Pith},
  title        = {Pith review of: Enhancing Visual Representation for Text-based Person Searching},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/MST32T3G}},
  note         = {Machine review of arXiv:2412.20646}
}
abstract

Text-based person search aims to retrieve the matched pedestrians from a large-scale image database according to the text description. The core difficulty of this task is how to extract effective details from pedestrian images and texts, and achieve cross-modal alignment in a common latent space. Prior works adopt image and text encoders pre-trained on unimodal data to extract global and local features from image and text respectively, and then global-local alignment is achieved explicitly. However, these approaches still lack the ability of understanding visual details, and the retrieval accuracy is still limited by identity confusion. In order to alleviate the above problems, we rethink the importance of visual features for text-based person search, and propose VFE-TPS, a Visual Feature Enhanced Text-based Person Search model. It introduces a pre-trained multimodal backbone CLIP to learn basic multimodal features and constructs Text Guided Masked Image Modeling task to enhance the model's ability of learning local visual details without explicit annotation. In addition, we design Identity Supervised Global Visual Feature Calibration task to guide the model learn identity-aware global visual features. The key finding of our study is that, with the help of our proposed auxiliary tasks, the knowledge embedded in the pre-trained CLIP model can be successfully adapted to text-based person search task, and the model's visual understanding ability is significantly enhanced. Experimental results on three benchmarks demonstrate that our proposed model exceeds the existing approaches, and the Rank-1 accuracy is significantly improved with a notable margin of about $1\%\sim9\%$. Our code can be found at https://github.com/zhangweifeng1218/VFE_TPS.

Figures

Figures reproduced from arXiv: 2412.20646 by the authors.

Figure 1
Figure 1. (a) Person identity confusion: The pedestrians with same ID have lower similarity [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. The overall framework of our Visual Feature Enhanced Text-based Person Search [PITH_FULL_IMAGE:figures/full_fig_p010_2.png] view at source ↗
Figure 3
Figure 3. Illustration of our TG-MIM and the popular SimMIM. (a) SimMIM predicts raw [PITH_FULL_IMAGE:figures/full_fig_p013_3.png] view at source ↗
Figures from the paper (6 more)
Figure 4
Figure 4. Figure 4: Illustration of our Identity Supervised Global Visual Feature Calibration (IS-GVFC). [PITH_FULL_IMAGE:figures/full_fig_p016_4.png]
Figure 5
Figure 5. Figure 5: Ablation study of different masked image modeling methods. [PITH_FULL_IMAGE:figures/full_fig_p028_5.png]
Figure 6
Figure 6. Figure 6: Ablation study of masked image modeling. (a) mAP curves of MAE, SimMIM [PITH_FULL_IMAGE:figures/full_fig_p029_6.png]
Figure 7
Figure 7. Figure 7: Ablation study of global visual feature calibration. [PITH_FULL_IMAGE:figures/full_fig_p030_7.png]
Figure 8
Figure 8. Figure 8: Visualization of global visual features under different calibration methods. Each [PITH_FULL_IMAGE:figures/full_fig_p031_8.png]
Figure 9
Figure 9. Figure 9: Visualization of searching results. The first row of each sample shows the returned [PITH_FULL_IMAGE:figures/full_fig_p033_9.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

56 extracted references · 50 canonical work pages

  1. [1]

    S. Li, T. Xiao, H. Li, Person search with natural language description, in: IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2017, pp. 1970–1979

  2. [2]

    J. Yu, W. Zhang, Y. Lu, Z. Qin, Y. Hu, J. Tan, Q. Wu, Reasoning on the relation: Enhancing visual representation for visual question answering and cross-modal retrieval, IEEE Transactions on Multimedia 22 (12) (2020) 3196–3209

  3. [3]

    H. Wang, J. Shen, Y. Liu, Nformer: Robust person re-identification with neighbor transformer, in: IEEE Conference on Computer Vision and Pat- tern Recognition (CVPR), 2022, pp. 7287–7297

  4. [4]

    Gao, S.-M

    P. Gao, S.-M. Li, F. Gao, F. Wang, R.-Y. Yuan, H. Fujita, In defense and revival of bayesian filtering for thermal infrared object tracking, Knowledge- Based Systems 293 (2024) 111665

  5. [5]

    Gao, X.-Y

    P. Gao, X.-Y. Zhang, X.-L. Yang, F. Gao, H. Fujita, F. Wang, Robust visual tracking with extreme point graph-guided annotation: Approach and experiment, Expert Systems with Applications 238 (B) (2024) 122013

  6. [6]

    P. Gao, R. Yuan, F. Wang, L. Xiao, H. Fujita, Y. Zhang, Siamese atten- tional keypoint network for high performance visual tracking, Knowledge- based Systems 193 (2020) 105448

  7. [7]

    P. Gao, Y. Ma, K. Song, C. Li, F. Wang, L. Xiao, Y. Zhang, High perfor- mance visual tracking with circular and structural operators, Knowledge- based Systems 161 (2018) 240–253

  8. [8]

    Baltrusaitis, C

    T. Baltrusaitis, C. Ahuja, L.-P. Morency, Multimodal machine learning: A survey and taxonomy, IEEE Transactions on Multimedia 41 (2) (2019) 423–443. 35

Show all 56 references
  1. [9]

    K. Zhou, F. H. Hassan, G. Hoon, The state of the art for cross-modal retrieval: A survey, IEEE Access 11 (2023) 138568–138589

  2. [10]

    J. Zhou, B. Huang, W. Fan, Z. Cheng, Z. Zhao, W. Zhang, Text-based person search via local-relational-global fine grained alignment, Knowledge- Based Systems 262 (2023) 110–121

  3. [11]

    S. Li, T. Xiao, H. Li, B. Zhou, D. Yue, X. Wang, Person search with natural language description, in: IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2017, pp. 1970–1979

  4. [12]

    K. Niu, Y. Huang, L. Wang, Fusing two directions in cross-domain adap- tion for real life person search by language, in: The IEEE International Conference on Computer Vision Workshops, 2019, pp. 1815–1818

  5. [13]

    Zhang, H

    Y. Zhang, H. Lu, Deep cross-modal projection learning for image-text matching, in: European Conference on Computer Vision (ECCV), 2018, pp. 686–701

  6. [14]

    Y. Chen, G. Zhang, Y. Lu, Z. Wang, Y. Zheng, R. Wang, TIPCB: A simple but effective part-based convolutional baseline for text-based person search, Neurocomputing 494 (2022) 171–181

  7. [15]

    Y. Jing, C. Si, J. Wang, W. Wang, L. Wang, T. Tan, Pose-guided multi- granularity attention network for text-based person search, in: AAAI Con- ference on Artificial Intelligence (AAAI), 2020, pp. 11189–11196

  8. [16]

    Z. Wang, Z. Fang, J. Wang, Y. Yang, VITAA: Visual-textual attributes alignment in person search by natural language, in: European Conference on Computer Vision (ECCV), 2020, pp. 402–420

  9. [17]

    L. Bao, L. Wei, W. Zhou, L. Liu, L. Xie, H. Li, Q. Tian, Multi-granularity matching transformer for text-based person search, IEEE Transactions on Multimedia 26 (2) (2024) 4281–4293. 36

  10. [18]

    Z. Wang, Z. Fang, J. Wang, Y. Yang, Beyond part models: Person retrieval with refined part pooling and a strong convolutional baseline, in: European Conference on Computer Vision (ECCV), 2018, pp. 480–496

  11. [19]

    K. Sun, B. Xiao, D. Liu, J. Wang, Deep high-resolution representation learning for human pose estimation, in: IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2019, pp. 5693–5703

  12. [20]

    Klein, C

    D. Klein, C. Manning, Fast exact inference with a factored model for nat- ural language parsing, in: Advances in Neural Information Processing Sys- tems (NIPS), 2002, pp. 3–10

  13. [21]

    Radford, J

    A. Radford, J. W. Kim, C. Hallacy, A. Ramesh, G. Goh, S. Agarwal, G. Sastry, A. Askell, P. Mishkin, J. Clark, Learning transferable visual models from natural language supervision, in: International Conference on Machine Learning (ICML), 2021, pp. 8748–8763

  14. [22]

    P. Vasu, H. Pouransari, F. Faghri, R. Vemulapalli, Mobileclip: Fast image- text models through multi-modal reinforced training, in: IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2024, pp. 369–378

  15. [23]

    M. Wang, J. Xing, B. Jiang, J. Chen, J. Mei, X. Zuo, G. Dai, J. Wang, Y. Liu, A multimodal, multi-task adapting framework for video action recognition, in: AAAI Conference on Artificial Intelligence (AAAI), 2024, pp. 5517–5525

  16. [24]

    X. Han, S. He, L. Zhang, T. Xiang, Text based person search with limited data, in: British Machine Vision Conference (BMVC), 2021, pp. 337–341

  17. [25]

    S. Yan, N. Dong, L. Zhang, Clip-driven fine-grained text-image person re-identification, IEEE Transactions on Image Processing 32 (2023) 6032– 6046

  18. [26]

    K. He, X. Chen, S. Xie, Y. Li, P. Dollar, R. Girshick, Masked autoencoders are scalable vision learners, in: IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2022, pp. 15979–15988. 37

  19. [27]

    Z. Xie, Z. Zhang, Y. Cao, SimMIM: a simple framework for masked image modeling, in: IEEE Conference on Computer Vision and Pattern Recogni- tion (CVPR), 2022, pp. 9643–9653

  20. [28]

    S. Li, T. Xiao, H. Li, W. Yang, X. Wang, Identity-aware textual-visual matching with latent co-attention, in: The IEEE International Conference on Computer Vision (ICCV), 2017, pp. 1890–1899

  21. [29]

    Simonyan, A

    K. Simonyan, A. Zisserman, Very deep convolutional networks for large- scale image recognition, in: arXiv preprint arXiv:1409.1556, 2014

  22. [30]

    Zheng, L

    Z. Zheng, L. Zheng, M. Garrett, Y. Yang, S. Y. D. Xu, M., Dual-path con- volutional image-text embeddings with instance loss, ACM Transactions on Multimedia Computing, Communications, and Applications 16 (2) (2022) 1–23

  23. [31]

    K. Niu, Y. Huang, W. Yang, L. Wang, Improving description-based person re-identification by multi-granularity image-text alignments, IEEE Trans- actions on Image Processing 29 (2020) 5542–5556

  24. [32]

    Z. Ding, C. Ding, Z. Shao, Semantically self-aligned network for text-to- image part-aware person re-identification, in: arXiv:2107.12666, 2021

  25. [33]

    A. Zhu, Z. Wang, Y. Li, et al, DSSL: Deep surroundings-person separation learning for text-based person retrieval, in: ACM International Conference on Multimedia, 2021, pp. 209–217

  26. [34]

    C. Gao, G. Cai, X. Jiang, et al, Contextual non-local alignment over full- scale representation for text-based person search, in: arXiv:2101.03036, 2021

  27. [35]

    Sarafianos, X

    N. Sarafianos, X. Xu, I. Kakadiaris, Adversarial representation learning for text-to-image matching, in: The IEEE International Conference on Computer Vision (ICCV), 2019, pp. 5814–5824. 38

  28. [36]

    Vaswani, N

    A. Vaswani, N. Shazeer, N. Parmar, J. Uszkoreit, L. Jones, et al, Attention is all you need, in: Advances in neural information processing systems (NIPS), 2017, pp. 5998–6008

  29. [37]

    S. Yan, N. Dong, L. Zhang, J. Tang, Clip-driven fine-grained text-image person re-identification, IEEE Transactions on Image Processing 32 (2023) 6032–6046

  30. [38]

    Devlin, M.-W

    J. Devlin, M.-W. Chang, L. K. Toutanova, Bert: Pre-training of deep bidi- rectional transformers for language understanding, in: Annual Conference of the North American Chapter of the Association for Computational Lin- guistics: Human LanguageTechnologies, 2019, pp. 4171–4186

  31. [39]

    D. Wei, S. Zhang, T. Yang, Y. Liu, J. Liu, Calibrating cross-modal features for text-based person searching, in: arXiv:2304.02278, 2023

  32. [40]

    Y. Bai, M. Cao, D. Gao, Z. Cao, C. Chen, Z. Fan, L. Nie, M. Zhang, Rasa: Relation and sensitivity aware representation learning for text-based person search, in: International Joint Conference on Artificial Intelligence (IJCAI), 2023, pp. 555–563

  33. [41]

    Jiang, M

    D. Jiang, M. Ye, Cross-modal implicit relation reasoning and aligning for text-to-image person retrieval, in: IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2023, pp. 2787–2797

  34. [42]

    Sennrich, B

    R. Sennrich, B. Haddow, A. Birch, Neural machine translation of rare words with subword units, in: Annual Meeting of the Association for Computa- tional Linguistics (ACL), 2016, pp. 523–536

  35. [43]

    W. Shi, J. Caballero, F. Huszar, J. Totz, A. Aitken, R. Bishop, D. Rueck- ert, Z. Wang, Real-time single image and video super-resolution using an efficient sub-pixel convolutional neural network, in: IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2016, pp....

  36. [44]

    W. Suo, M. Sun, K. Niu, et al, A simple and robust correlation filtering method for text-based person search, in: European Conference on Com- puter Vision (ECCV), 2022, pp. 726–742

  37. [45]

    S. Li, M. Cao, M. Zhang, Learning semantic-aligned feature representa- tion for text-based person search, in: IEEE International Conference on Acoustics, Speech and Signal Processing, 2022, pp. 2724–2728

  38. [46]

    Farooq, M

    A. Farooq, M. Awais, J. Kittler, et al, Axm-net: Implicit cross-modal feature alignment for person re-identification, in: AAAI Conference on Artificial Intelligence (AAAI), 2022, pp. 4477–4485

  39. [47]

    S. Yan, H. Tang, L. Zhang, J. Tang, Learning granularity-unified repre- sentations for text-to-image person reidentification, in: ACM International Conference on Multimedia, 2022, pp. 425–436

  40. [48]

    X. Shu, W. Wen, H. Wu, et al, See finer, see more: Implicit modality align- ment for text-based person retrieval, in: European Conference on Computer Vision (ECCV), 2022, pp. 624–641

  41. [49]

    Y. Liu, Y. Li, Z. Liu, W. Yang, Y. Wang, Q. Liao, Clip-based synergistic knowledge transfer for text-based person retrieval, in: arXiv:2309.09496v1, 2023

  42. [50]

    G. Wang, F. Yu, J. Li, Q. Jia, S. Ding, Exploiting the textual poten- tial from vision-language pre-training for text-based person search, in: arXiv:2303.04497, 2023

  43. [51]

    S. He, H. Luo, W. Jiang, et al, Vgsg: Vision-guided semantic-group network for text-based person search, IEEE Transactions on Image Processing 33 (2024) 163–176

  44. [52]

    S. Yan, H. Tang, L. Zhang, J. Tang, Image-specific information sup- pression and implicit local alignment for text-based person search, in: arXiv:2208.14365, 2022. 40

  45. [53]

    Z. Wang, A. Zhu, J. Xue, et al, Look before you leap: Improving text-based person retrieval by learning a consistent cross-modal common manifold, in: ACM International Conference on Multimedia, 2022, pp. 1984–1992

  46. [54]

    J. Li, R. R. Selvaraju, A. Gotmare, S. R. Joty, C. Xiong, S. C.-H. Hoi, Align before fuse: Vision and language representation learning with mo- mentum distillation, in: Advances in Neural Information Processing Sys- tems (NeurlPS), 2021, pp. 9694–9705

  47. [55]

    Bagirov, R

    A. Bagirov, R. Aliguliyev, N. Sultanova, Finding compact and well- separated clusters: Clustering using silhouette coefficients, Pattern Recog- nition 135 (2023) 109–144

  48. [56]

    Laurens, G

    M. Laurens, G. Hinton, Visualizing data using t-sne, Journal of Machine Learning Research 9 (1) (2008) 2579–2605. 41

Pith tools

Reviewed August 10, 2026 · model on record in the stance chip above.