Pith. sign in

REVIEW 3 major objections 5 minor 59 references

Learning Semantic-Aware Representation in Visual-Language Models for Multi-Label Recognition with Partial Labels

T0 review · 3 major / 5 minor · reviewed 2026-08-11 · deepseek-v4-flash

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

Pith's one-line read A frozen CLIP visual encoder followed by a small semantic-decoupling attention module and per-category learned prompts attains state-of-the-art multi-label recognition with partial labels.

desk verdict Strong empirical gains on CLIP-based partial-label multi-label recognition, but the independence claim is contradicted by the softmax in Eq. 9; worth a referee who insists on a fix. read the letter →

arxiv 2412.10843 v1 pith:CPWQC3HJ submitted 2024-12-14 cs.CV

Haoxian Ruan , Zhihua Xu , Zhijing Yang , Yongyi Lu , Jinghui Qin , Tianshui Chen This is my paper · ORCID
classification cs.CV
keywords multi-labelrecognitionpartiallabelsvision-languagemodelsCLIPpromptlearningsemanticdecouplingspatialattentionasymmetricloss
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

The paper claims that the obstacle to using CLIP for multi-label recognition with partial labels is semantic confusion: one global image representation and one shared text prompt entangle categories that can co-occur. To remove that obstacle, it adds a small trainable semantic-decoupling module after the frozen CLIP visual encoder that uses category semantics to attend to category-specific regions, and it learns a separate prompt per category through the frozen text encoder. On MS-COCO and Pascal VOC with 10-90% of labels known, the framework reports average mAP of 81.7 and 92.8, exceeding the best previous methods by 3.3 and 1.3 points. If correct, partial-label multi-label recognition can be handled by a simple, parameter-efficient CLIP adaptation rather than by complex label-correlation models.

What carries the argument

The load-bearing object is the semantic decoupling module: for each category $c$, it takes the CLIP visual feature map $f^g$ and a GloVe semantic vector $x_c$, fuses them at every spatial location with low-rank bilinear pooling $\tilde{f}_{c,wh}=P^T(\tanh((U^T f^g_{wh})\odot(V^T x_c)))+b$, normalizes the resulting attention scores $a_{c,wh}$ over all spatial locations, and pools the visual features into a category-specific vector $f^v_c$. On the text side, each category has its own prompt $t_c=[V]^c_1\cdots[V]^c_M[CLS]$ whose tokens are learned while the text encoder stays frozen. The prediction score for category $c$ is the softmax-normalized cosine similarity $\exp(\langle f^v_c, f^t_c\rangle/\tau)$ over all categories, and training uses the partial asymmetric loss (P-ASL) on known labels only. Only the decoupling module and the prompt tokens are updated; the CLIP encoders stay fixed.

What would settle it

Retrain the framework on MS-COCO with 10% known labels after replacing the category softmax in Eq. (9) with independent per-category sigmoid scores; if mAP does not fall below the reported 78.3, the independence mechanism claimed by the paper is not what carries the improvement.

Watch

Extended reading notes

Core claim

The paper's central claim is that the semantic confusion that limits CLIP on multi-label recognition with partial labels can be resolved by decoupling rather than by enlarging the model. A semantic-guided spatial attention module produces a separate visual feature per category from the frozen CLIP feature map, and category-specific learnable prompts produce matching text features, so each category is scored from its own visual-text pair. With only the decoupling module and prompt tokens trainable, the framework reaches average mAP of 81.7 on MS-COCO and 92.8 on Pascal VOC across 10-90% known labels and outperforms all compared methods, using about 4.8 million trainable parameters. The ablation study attributes the larger share of the gain to the decoupling module (5.5 mAP over the prompt-only baseline on COCO), and the limitation section reports remaining failure cases for small or heavily occluded objects whose attention maps become inaccurate or over-broad.

Load-bearing premise

The load-bearing premise is that scoring each category with a softmax over all categories still lets each category be treated independently, even though the softmax denominator couples all category scores; if that coupling materially hurts multi-label prediction, the claim that per-category decoupling is what drives the gains is undermined.

Editorial extensions

If this is right

  • If the reported COCO gains hold, 10% known labels are enough to reach 78.3 mAP, roughly matching what complete-label methods needed far more supervision to achieve.
  • The frozen-encoder design means adapting to a new label set costs only training the small decoupling module and prompt tokens, with no full-model fine-tuning.
  • Because category-specific prompts improve the baseline even without decoupling (76.9 versus 74.6 average mAP on COCO), prompt specialization is a transferable improvement for CLIP-based multi-label recognition.
  • The P-ASL loss contributes most when labels are scarce, below 30% known labels, so loss choice is part of the claimed gains in the hardest partial-label settings.
  • The attention maps that separate co-occurring categories imply the decoupling step also provides implicit object localization, not just classification scores.

Reading between the lines

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

  • I infer the same decoupling module could be grafted onto larger or newer frozen vision-language encoders, since it adds only a few million parameters and does not retrain the base encoders; the paper only tests a ResNet-101 CLIP.
  • The paper leaves label-correlation modeling as future work, and adding co-occurrence or graph-based context to the decoupled features could address the small-object and occlusion failures reported in its limitation section.
  • A direct testable extension would be to apply the category-specific prompt component in a zero-shot multi-label setting without training labels, to isolate how much of the gain comes from prompt specialization versus the decoupling module.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

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 a CLIP-based framework for multi-label recognition with partial labels (MLR-PL). The method freezes the CLIP visual and textual encoders and trains two lightweight components: (i) a semantic decoupling module that uses GloVe semantic embeddings and low-rank bilinear pooling to compute category-specific spatial attention maps, and (ii) a set of independent, learnable category-specific prompts (following CoOp) that produce per-category text embeddings. Predictions are obtained by matching decoupled visual features with category-specific text features, and training uses a partial asymmetric loss (P-ASL) on the known labels only. Experiments on MS-COCO 2014 and Pascal VOC 2007 across 10%–90% known label proportions report consistent improvements in mAP, OF1, and CF1 over several prior MLR-PL methods, with an average mAP gain of 3.3 points on COCO and 1.3 points on VOC over the best baseline. The paper also includes ablations, per-class AP analysis, and visualizations of attention maps.

Significance. If the reported gains hold, the contribution is practically useful: it shows that a frozen CLIP backbone with a small trainable attention module and per-category prompts can achieve strong MLR-PL performance with a simpler architecture than prior label-correlation models. The paper is generally careful on the experimental side: it reports error bars, runs ablations that isolate the semantic decoupling module and category-specific prompts, provides per-class AP breakdowns, and includes failure-case analysis. However, the central mechanistic claim of the paper — that category predictions are independent because of the decoupling design — is directly contradicted by the softmax normalization in Eq. (9), and the state-of-the-art comparison omits recent CLIP-based multi-label methods, which are the most natural competitors. These two issues need to be resolved before the claims can be accepted as stated.

major comments (3)
  1. [§3.2, Eq. (9); Abstract; §3 intro] The paper repeatedly states that 'the prediction of each category is independent' (Abstract and §3), and this independence is the stated motivation for the semantic decoupling module. However, Eq. (9) computes p_c = exp(<f^v_c, f^t_c>/τ) / Σ_{c'=1}^C exp(<f^v_{c'}, f^t_{c'}>/τ), which is a softmax over all C categories. In a multi-label image, several categories are simultaneously present; under this formulation, increasing the score of one present category necessarily decreases the scores of all other categories, including other present ones. The P-ASL loss in Eq. (10) is then applied to these coupled probabilities, so the training signal for one positive category pushes against other positive categories. This directly undermines the independence claim that motivates the architecture. Please either replace the softmax with per-category sigmoid or logistic normalization (e.g., p_c = sigmoid(<f^v_c, f^t_c>/τ)), which would make predictions truly independent, or explicitly justify why a shared-denominator softmax is appropriate for multi-label scoring and reconcile that with the independence claim. As written, the model is a single softmax classifier over categories, not C independent classifiers.
  2. [§4.5, Fig. 8; §4.2] The hyperparameters γ+ and γ− of P-ASL are tuned by scanning mAP on MS-COCO at 50% and 80% known-label settings (Fig. 8), and the same values (γ+ = 1, γ− = 2) are then used to produce the final reported results on MS-COCO and Pascal VOC (Table 2). This means the final evaluation on MS-COCO is conducted on a dataset that was already used for hyperparameter selection, which is a mild form of selection on the evaluation metric. Please report results with hyperparameters fixed by a separate validation split, or provide a sensitivity analysis showing that the performance gap is robust to the choice of γ+ and γ− on both datasets. This is particularly important because the advantage over the best baseline at high label proportions is small on VOC (1.3 points average mAP).
  3. [§4.3, Table 2] The state-of-the-art comparison includes graph-based methods (SSGRL, GCN-ML, KGGR), pseudo-label methods (SST, HST), and feature-blending methods (SARB, DSRB), but it does not include any recent CLIP-based multi-label classification methods. Since the proposed method is built on CLIP and the paper's motivation is that CLIP-based methods suffer from semantic confusion, the claim that the method 'significantly outperforms current state-of-art methods' needs to be tested against other CLIP-based approaches, such as CLIP zero-shot, CoOp, CoCoOp, and any CLIP-based MLR methods published in 2023–2024. The CoOp baseline appears only in the ablation table (Table 4), not in the main comparison, so the reader cannot see how the proposed method compares to the most direct CLIP-based competitor on the same evaluation protocol. Please add these comparisons or justify why they are excluded.
minor comments (5)
  1. [§3.1, Eq. (6)] Eq. (6) uses f^g_{c,wh} but the global feature at position (w,h) is defined as f^g_{wh} in Eq. (1) and Eq. (3); the subscript c is not defined for the global feature and appears to be a typo, since the attention pooling sums over spatial locations with the same global feature for all categories.
  2. [Fig. 6] The word 'toile' appears in the predictions; this should be 'toilet'.
  3. [Abstract] 'which alleviate the semantic confusion problem' should be 'which alleviates the semantic confusion problem' to agree with the singular subject 'prediction of each category is independent'.
  4. [§4.3, Table 2] The tables would be easier to interpret if the second-best baseline were marked or if the average-gain sentence explicitly said which method is the second-best; currently Table 2 shows DSRB as best on VOC and HST/DSRB on COCO, but the text does not name them consistently.
  5. [§3.3, Eq. (10)] The definition of p̄^n_c = max(p^n_c − m, 0) uses the hard threshold m for negative samples, but the paper does not discuss whether this threshold interacts with the softmax normalization in Eq. (9); a brief comment on how m is chosen and how it behaves under the shared denominator would be helpful.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: reported gains are measured on held-out complete test labels, and the cited prior works are baselines rather than load-bearing justifications.

full rationale

The central claim is an empirical performance comparison, not a derivation from first principles. Training supervises the learnable semantic-decoupling module and category-specific prompts with P-ASL on known labels only (Eqs. 1-11), and evaluation uses completely annotated test sets on MS-COCO and Pascal VOC, so the reported mAP figures are not equivalent to the training inputs by construction. The method is anchored externally: the prompt-optimization component is adopted from CoOp [53], Table 4 ablates the two proposed modules against that external baseline, and Tables 2-3 compare against independent and author-group baselines. Author self-citations (e.g., SST, HST, SARB, DSRB, KGGR, SSGRL) are used as related work and comparison methods, not as justification for the proposed mechanism, so no load-bearing self-citation chain exists. Two legitimate concerns remain but are not circularity: Sec. 4.5 selects the P-ASL hyperparameters gamma+ and gamma- by sweeping mAP on MS-COCO and reuses them on the same benchmark, a mild selection-on-metric risk; and Eq. (9) normalizes scores with a softmax over all C categories, so category predictions are in fact coupled despite the abstract's independence claim. Both are correctness or reporting concerns that do not make any predicted quantity equal to its input by definition.

Assumptions & free parameters 4 free parameters · 5 assumptions · 0 invented entities

The method adds two learnable components, an attention module and per-category prompts, plus several tuned hyperparameters. No new physical or conceptual entity is introduced. The strongest implicit assumption is that GloVe semantics guide CLIP spatial features; the most inconsistent modeling choice is softmax normalization for a multi-label task.

free parameters (4)
  • gamma+ (P-ASL focusing for positives) = 1
    Tuned on MS-COCO at 50% and 80% known labels in Fig. 8; affects the loss weighting for positive samples.
  • gamma- (P-ASL focusing for negatives) = 2
    Tuned on MS-COCO in Fig. 8; strongly affects mAP, which rises from 76.5 to 82.0 as gamma- increases to 2.
  • hard margin m for negative probabilities = 0.05
    Set in implementation; down-weights easy negative samples in P-ASL.
  • prompt length M = 16 tokens
    Chosen for category-specific prompts; no ablation is provided for this choice.
assumptions (5)
  • domain assumption CLIP's pretrained visual and textual encoders, kept frozen, provide representations transferable to MLR-PL.
    The whole framework relies on frozen CLIP features being rich enough for the partial-label task; invoked in Section 3.
  • domain assumption GloVe category-name embeddings align sufficiently with CLIP visual feature space to guide spatial attention.
    Equations (2)-(3) use GloVe vectors as semantic guidance; if this alignment is poor, the attention maps would not locate category-relevant regions.
  • domain assumption Randomly dropping known labels from fully annotated COCO and VOC datasets simulates real partial-label conditions.
    Section 4.1 creates partial labels by random dropping; realistic partial annotations may have systematic missing-label patterns.
  • ad hoc to paper Softmax over categories in Eq. (9) is a valid way to score positives in a multi-label task.
    Multi-label categories are not mutually exclusive, yet predictions are normalized across all categories; this contradicts the stated independence of category predictions.
  • domain assumption P-ASL loss computed on annotated samples only is appropriate for the partial-label setting.
    Equation (10) applies the asymmetric loss to known labels only; the method assumes this handles the positive-negative imbalance under partial supervision.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Learning Semantic-Aware Representation in Visual-Language Models for Multi-Label Recognition with Partial Labels." pith.science (2026). https://pith.science/paper/CPWQC3HJ

@misc{pith2026241210843,
  author       = {Pith},
  title        = {Pith review of: Learning Semantic-Aware Representation in Visual-Language Models for Multi-Label Recognition with Partial Labels},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/CPWQC3HJ}},
  note         = {Machine review of arXiv:2412.10843}
}
read the original abstract

Multi-label recognition with partial labels (MLR-PL), in which only some labels are known while others are unknown for each image, is a practical task in computer vision, since collecting large-scale and complete multi-label datasets is difficult in real application scenarios. Recently, vision language models (e.g. CLIP) have demonstrated impressive transferability to downstream tasks in data limited or label limited settings. However, current CLIP-based methods suffer from semantic confusion in MLR task due to the lack of fine-grained information in the single global visual and textual representation for all categories. In this work, we address this problem by introducing a semantic decoupling module and a category-specific prompt optimization method in CLIP-based framework. Specifically, the semantic decoupling module following the visual encoder learns category-specific feature maps by utilizing the semantic-guided spatial attention mechanism. Moreover, the category-specific prompt optimization method is introduced to learn text representations aligned with category semantics. Therefore, the prediction of each category is independent, which alleviate the semantic confusion problem. Extensive experiments on Microsoft COCO 2014 and Pascal VOC 2007 datasets demonstrate that the proposed framework significantly outperforms current state-of-art methods with a simpler model structure. Additionally, visual analysis shows that our method effectively separates information from different categories and achieves better performance compared to CLIP-based baseline method.

Figures

Figures reproduced from arXiv: 2412.10843 by the authors.

Figure 1
Figure 1. An illustration of MLR image with complete labels [a] and partial labels [b]. All positive and negative [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. An overall illustration of the proposed framework. For each class, a prompt is initialized with each [PITH_FULL_IMAGE:figures/full_fig_p006_2.png] view at source ↗
Figure 3
Figure 3. Several examples of input images and category activation maps corresponding to the top-3 highest [PITH_FULL_IMAGE:figures/full_fig_p011_3.png] view at source ↗
Figures from the paper (6 more)
Figure 4
Figure 4. Figure 4: Comparison of the proposed semantic decoupling module and semantic attention fusion method (SA) [PITH_FULL_IMAGE:figures/full_fig_p012_4.png]
Figure 5
Figure 5. Figure 5: Per-class average precision (AP) of our proposed framework and baseline method with known label [PITH_FULL_IMAGE:figures/full_fig_p013_5.png]
Figure 6
Figure 6. Figure 6: Visualizations of multi-label classification results for several images predicted by our proposed method [PITH_FULL_IMAGE:figures/full_fig_p014_6.png]
Figure 7
Figure 7. Figure 7: Comparison of P-ASL and P-BCE as loss functions for the proposed framework. All the experiments [PITH_FULL_IMAGE:figures/full_fig_p014_7.png]
Figure 8
Figure 8. Figure 8: Analysis of the effect of 𝛾+ (left) and 𝛾− (right) in P-ASL. These experiments are conducted on the MS-COCO dataset with known label proportions of 50% and 80%. Moreover, the parameters 𝛾+ and 𝛾− are critical to the performance of P-ASL. To gain a more comprehensive un…
Figure 9
Figure 9. Figure 9: Examples of failure cases in the proposed semantic decoupling method. Typical failures include [PITH_FULL_IMAGE:figures/full_fig_p015_9.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

59 extracted references · 41 canonical work pages

  1. [1]

    Adrian Bulat and Georgios Tzimiropoulos. 2023. Lasp: Text-to-text optimization for language-aware soft prompting of vision & language models. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition . 23232–23241

    work page 2023

  2. [2]

    Dolly Carrillo, Vivian F López, and María N Moreno. 2013. Multi-label classification for recommender systems. In Trends in Practical Applications of Agents and Multiagent Systems: 11th International Conference on Practical Applications of Agents and Multi-Agent Systems . Springer, 181–188

    work page 2013

  3. [3]

    Tianshui Chen, Liang Lin, Riquan Chen, Xiaolu Hui, and Hefeng Wu. 2022. Knowledge-Guided Multi-Label Few-Shot Learning for General Image Recognition. IEEE Transactions on Pattern Analysis and Machine Intelligence 44, 3 (2022), 1371–1384. https://doi.org/10.1109/TPAMI.2020.3025814

    arXiv 2022

  4. [4]

    Tianshui Chen, Tao Pu, Lingbo Liu, Yukai Shi, Zhijing Yang, and Liang Lin. 2024. Heterogeneous semantic transfer for multi-label recognition with partial labels. International Journal of Computer Vision (2024), 1–16

    work page 2024

  5. [5]

    Tianshui Chen, Tao Pu, Hefeng Wu, Yuan Xie, and Liang Lin. 2022. Structured semantic transfer for multi-label recognition with partial labels. In Proceedings of the AAAI conference on artificial intelligence , Vol. 36. 339–346. ACM Trans. Multimedia Comput. Commun. Appl., Vol. 1, No. 1, Article . Publication date: December 2024. Learning Semantic-Aware Re...

    work page 2022

  6. [6]

    Tianshui Chen, Tao Pu, Hefeng Wu, Yuan Xie, Lingbo Liu, and Liang Lin. 2021. Cross-domain facial expression recognition: A unified evaluation benchmark and adversarial graph learning. IEEE transactions on pattern analysis and machine intelligence 44, 12 (2021), 9887–9903

    work page 2021

  7. [7]

    Tianshui Chen, Weihang Wang, Tao Pu, Jinghui Qin, Zhijing Yang, Jie Liu, and Liang Lin. 2024. Dynamic Correlation Learning and Regularization for Multi-Label Confidence Calibration. arXiv preprint arXiv:2407.06844 (2024)

    work page Pith review arXiv 2024

  8. [8]

    Tianshui Chen, Zhouxia Wang, Guanbin Li, and Liang Lin. 2018. Recurrent Attentional Reinforcement Learning for Multi-Label Image Recognition. Proceedings of the AAAI Conference on Artificial Intelligence 32, 1 (Apr. 2018). https://doi.org/10.1609/aaai.v32i1.12281

    work page doi:10.1609/aaai.v32i1.12281 2018

Show all 59 references
  1. [9]

    Tianshui Chen, Muxin Xu, Xiaolu Hui, Hefeng Wu, and Liang Lin. 2019. Learning Semantic-Specific Graph Represen- tation for Multi-Label Image Recognition. In Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV)

    2019

  2. [10]

    Zhao-Min Chen, Xiu-Shen Wei, Peng Wang, and Yanwen Guo. 2019. Multi-label image recognition with graph convolutional networks. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition . 5177– 5186

    2019

  3. [11]

    Shyi-Chyi Cheng, Tzu-Chuan Chou, Chao-Lung Yang, and Hung-Yi Chang. 2005. A semantic learning for content-based image retrieval using analytical hierarchy process. Expert Systems with Applications 28, 3 (2005), 495–505

    2005

  4. [12]

    Zahra Zamanzadeh Darban and Mohammad Hadi Valipour. 2022. GHRS: Graph-based hybrid recommendation system with application to movie recommendation. Expert Systems with Applications 200 (2022), 116850

    2022

  5. [13]

    Alexey Dosovitskiy, Lucas Beyer, Alexander Kolesnikov, Dirk Weissenborn, Xiaohua Zhai, Thomas Unterthiner, Mostafa Dehghani, Matthias Minderer, Georg Heigold, Sylvain Gelly, Jakob Uszkoreit, and Neil Houlsby. 2021. An Image is Worth 16x16 Words: Transformers for Image Recognit...

    2021

  6. [14]

    Yu Du, Fangyun Wei, Zihe Zhang, Miaojing Shi, Yue Gao, and Guoqi Li. 2022. Learning to prompt for open-vocabulary object detection with vision-language model. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 14084–14093

    2022

  7. [15]

    Thibaut Durand, Nazanin Mehrasa, and Greg Mori. 2019. Learning a deep convnet for multi-label classification with partial labels. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition . 647–657

    2019

  8. [16]

    Mark Everingham, Luc Van Gool, Christopher KI Williams, John Winn, and Andrew Zisserman. 2010. The pascal visual object classes (voc) challenge. International journal of computer vision 88 (2010), 303–338

    2010

  9. [17]

    Chunjiang Ge, Rui Huang, Mixue Xie, Zihang Lai, Shiji Song, Shuang Li, and Gao Huang. 2023. Domain Adaptation via Prompt Learning. IEEE Transactions on Neural Networks and Learning Systems (2023), 1–11. https://doi.org/10. 1109/TNNLS.2023.3327962

    2023

  10. [18]

    Kaiming He, Xiangyu Zhang, Shaoqing Ren, and Jian Sun. 2016. Deep residual learning for image recognition. In Proceedings of the IEEE conference on computer vision and pattern recognition . 770–778

    2016

  11. [19]

    Sunan He, Taian Guo, Tao Dai, Ruizhi Qiao, Xiujun Shu, Bo Ren, and Shu-Tao Xia. 2023. Open-Vocabulary Multi-Label Classification via Multi-Modal Knowledge Transfer. Proceedings of the AAAI Conference on Artificial Intelligence 37, 1 (Jun. 2023), 808–816. https://doi.org/10.160...

    2023 doi

  12. [20]

    Sepp Hochreiter and Jürgen Schmidhuber. 1997. Long short-term memory. Neural computation 9, 8 (1997), 1735–1780

    1997

  13. [21]

    Dat Huynh and Ehsan Elhamifar. 2020. Interactive multi-label cnn learning with partial labels. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition . 9423–9432

    2020

  14. [22]

    Chao Jia, Yinfei Yang, Ye Xia, Yi-Ting Chen, Zarana Parekh, Hieu Pham, Quoc Le, Yun-Hsuan Sung, Zhen Li, and Tom Duerig. 2021. Scaling Up Visual and Vision-Language Representation Learning With Noisy Text Supervision. In Proceedings of the 38th International Conference on Mach...

    2021

  15. [23]

    Menglin Jia, Luming Tang, Bor-Chun Chen, Claire Cardie, Serge Belongie, Bharath Hariharan, and Ser-Nam Lim

  16. [24]

    Dahun Kim, Anelia Angelova, and Weicheng Kuo. 2023. Region-aware pretraining for open-vocabulary object detection with vision transformers. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition . 11144–11154

    2023

  17. [25]

    Jin-Hwa Kim, Kyoung-Woon On, Woosang Lim, Jeonghee Kim, Jung-Woo Ha, and Byoung-Tak Zhang. 2017. Hadamard Product for Low-rank Bilinear Pooling. In International Conference on Learning Representations . https://openreview. net/forum?id=r1rhWnZkg

    2017

  18. [26]

    Youngwook Kim, Jae Myung Kim, Zeynep Akata, and Jungwoo Lee. 2022. Large loss matters in weakly supervised multi-label classification. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition . 14156–14165. ACM Trans. Multimedia Comput. Commun. Appl...

    2022

  19. [27]

    Youngwook Kim, Jae Myung Kim, Jieun Jeong, Cordelia Schmid, Zeynep Akata, and Jungwoo Lee. 2023. Bridging the gap between model explanations in partially annotated multi-label classification. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition ...

    2023

  20. [28]

    Kipf and Max Welling

    Thomas N. Kipf and Max Welling. 2016. Semi-Supervised Classification with Graph Convolutional Networks. CoRR abs/1609.02907 (2016). arXiv:1609.02907 http://arxiv.org/abs/1609.02907

    2016 arXiv

  21. [29]

    Hanjiang Lai, Pan Yan, Xiangbo Shu, Yunchao Wei, and Shuicheng Yan. 2016. Instance-aware hashing for multi-label image retrieval. IEEE Transactions on Image Processing 25, 6 (2016), 2469–2479

    2016

  22. [30]

    Ran Li, YaFei Zhang, Zining Lu, Jianjiang Lu, and Yulong Tian. 2010. Technique of Image Retrieval Based on Multi-label Image Annotation. In 2010 Second International Conference on Multimedia and Information Technology , Vol. 2. 10–13. https://doi.org/10.1109/MMIT.2010.34

    2010 doi

  23. [31]

    Yujia Li, Daniel Tarlow, Marc Brockschmidt, and Richard Zemel. 2015. Gated graph sequence neural networks. arXiv preprint arXiv:1511.05493 (2015)

    2015 arXiv

  24. [32]

    Lawrence Zitnick

    Tsung-Yi Lin, Michael Maire, Serge Belongie, James Hays, Pietro Perona, Deva Ramanan, Piotr Dollár, and C. Lawrence Zitnick. 2014. Microsoft COCO: Common Objects in Context. In Computer Vision – ECCV 2014 , David Fleet, Tomas Pajdla, Bernt Schiele, and Tinne Tuytelaars (Eds.)....

    2014

  25. [33]

    Yuning Lu, Jianzhuang Liu, Yonggang Zhang, Yajing Liu, and Xinmei Tian. 2022. Prompt distribution learning. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition . 5206–5215

    2022

  26. [34]

    Adam Paszke, Sam Gross, Francisco Massa, Adam Lerer, James Bradbury, Gregory Chanan, Trevor Killeen, Zeming Lin, Natalia Gimelshein, Luca Antiga, Alban Desmaison, Andreas Kopf, Edward Yang, Zachary DeVito, Martin Raison, Alykhan Tejani, Sasank Chilamkurthy, Benoit Steiner, Lu ...

    2019

  27. [35]

    Jeffrey Pennington, Richard Socher, and Christopher D Manning. 2014. Glove: Global vectors for word representation. In Proceedings of the 2014 conference on empirical methods in natural language processing (EMNLP) . 1532–1543

    2014

  28. [36]

    Tao Pu, Tianshui Chen, Hefeng Wu, and Liang Lin. 2022. Semantic-aware representation blending for multi-label image recognition with partial labels. In Proceedings of the AAAI Conference on Artificial Intelligence , Vol. 36. 2091–2098

    2022

  29. [37]

    Tao Pu, Tianshui Chen, Hefeng Wu, Yukai Shi, Zhijing Yang, and Liang Lin. 2024. Dual-perspective semantic-aware representation blending for multi-label image recognition with partial labels. Expert Systems with Applications 249 (2024), 123526. https://doi.org/10.1016/j.eswa.20...

    2024

  30. [38]

    Tao Pu, Tianshui Chen, Yuan Xie, Hefeng Wu, and Liang Lin. 2021. AU-Expression Knowledge Constrained Represen- tation Learning for Facial Expression Recognition. In 2021 IEEE International Conference on Robotics and Automation (ICRA). 11154–11161. https://doi.org/10.1109/ICRA4...

    2021

  31. [39]

    Jie Qin, Jie Wu, Pengxiang Yan, Ming Li, Ren Yuxi, Xuefeng Xiao, Yitong Wang, Rui Wang, Shilei Wen, Xin Pan, et al

  32. [40]

    Alec Radford, Jong Wook Kim, Chris Hallacy, Aditya Ramesh, Gabriel Goh, Sandhini Agarwal, Girish Sastry, Amanda Askell, Pamela Mishkin, Jack Clark, Gretchen Krueger, and Ilya Sutskever. 2021. Learning Transferable Visual Models From Natural Language Supervision. In Proceedings...

    2021

  33. [41]

    Tal Ridnik, Emanuel Ben-Baruch, Nadav Zamir, Asaf Noy, Itamar Friedman, Matan Protter, and Lihi Zelnik-Manor

  34. [42]

    Haobo Wang, Ruixuan Xiao, Yixuan Li, Lei Feng, Gang Niu, Gang Chen, and Junbo Zhao. 2021. Pico: Contrastive label disambiguation for partial label learning. In International Conference on Learning Representations

    2021

  35. [43]

    Jiang Wang, Yi Yang, Junhua Mao, Zhiheng Huang, Chang Huang, and Wei Xu. 2016. Cnn-rnn: A unified framework for multi-label image classification. In Proceedings of the IEEE conference on computer vision and pattern recognition . 2285–2294

    2016

  36. [44]

    Shouwen Wang, Qian Wan, Xiang Xiang, and Zhigang Zeng. 2023. Saliency Regularization for Self-Training with Partial Annotations. In Proceedings of the IEEE/CVF International Conference on Computer Vision . 1611–1620

    2023

  37. [45]

    Zhouxia Wang, Tianshui Chen, Guanbin Li, Ruijia Xu, and Liang Lin. 2017. Multi-label image recognition by recurrently discovering attentional regions. In Proceedings of the IEEE international conference on computer vision . 464–472

    2017

  38. [46]

    Hefeng Wu, Yafei Hu, Keze Wang, Hanhui Li, Lin Nie, and Hui Cheng. 2019. Instance-aware representation learning and association for online multi-person tracking. Pattern Recognition 94 (2019), 25–34

    2019

  39. [47]

    Jin Ye, Junjun He, Xiaojiang Peng, Wenhao Wu, and Yu Qiao. 2020. Attention-driven dynamic graph convolutional network for multi-label image recognition. In Computer Vision–ECCV 2020: 16th European Conference, Glasgow, UK, ACM Trans. Multimedia Comput. Commun. Appl., Vol. 1, No...

    2020

  40. [48]

    Jin Yuan, Shikai Chen, Yao Zhang, Zhongchao Shi, Xin Geng, Jianping Fan, and Yong Rui. 2023. Graph Attention Transformer Network for Multi-label Image Classification. ACM Trans. Multimedia Comput. Commun. Appl. 19, 4, Article 150 (feb 2023), 16 pages. https://doi.org/10.1145/3578518

    2023 doi

  41. [49]

    Zhiwei Zhang and Hanyu Peng. 2021. Instance-weighted central similarity for multi-label image retrieval. arXiv preprint arXiv:2108.05274 (2021)

    2021 arXiv

  42. [50]

    Yong Zheng, Bamshad Mobasher, and Robin Burke. 2014. Context Recommendation Using Multi-label Classification. In 2014 IEEE/WIC/ACM International Joint Conferences on Web Intelligence (WI) and Intelligent Agent Technologies (IAT) , Vol. 2. 288–295. https://doi.org/10.1109/WI-IA...

    2014 doi

  43. [51]

    Yiwu Zhong, Jianwei Yang, Pengchuan Zhang, Chunyuan Li, Noel Codella, Liunian Harold Li, Luowei Zhou, Xiyang Dai, Lu Yuan, Yin Li, et al. 2022. Regionclip: Region-based language-image pretraining. In Proceedings of the IEEE/CVF conference on computer vision and pattern recogni...

    2022

  44. [52]

    Kaiyang Zhou, Jingkang Yang, Chen Change Loy, and Ziwei Liu. 2022. Conditional prompt learning for vision-language models. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition . 16816–16825

    2022

  45. [53]

    Kaiyang Zhou, Jingkang Yang, Chen Change Loy, and Ziwei Liu. 2022. Learning to prompt for vision-language models. International Journal of Computer Vision 130, 9 (2022), 2337–2348

    2022

  46. [54]

    Wei Zhou, Yanke Hou, Dihu Chen, Haifeng Hu, and Tao Su. 2023. Attention-Augmented Memory Network for Image Multi-Label Classification. ACM Trans. Multimedia Comput. Commun. Appl. 19, 3, Article 116 (feb 2023), 24 pages. https://doi.org/10.1145/3570166

    2023 doi

  47. [55]

    Wei Zhou, Zhiwu Xia, Peng Dou, Tao Su, and Haifeng Hu. 2023. Aligning Image Semantics and Label Concepts for Image Multi-Label Classification. ACM Trans. Multimedia Comput. Commun. Appl. 19, 2, Article 75 (feb 2023), 23 pages. https://doi.org/10.1145/3550278

    2023 doi

  48. [56]

    Wei Zhou, Zhiwu Xia, Peng Dou, Tao Su, and Haifeng Hu. 2023. Double Attention Based on Graph Attention Network for Image Multi-Label Classification. ACM Trans. Multimedia Comput. Commun. Appl. 19, 1, Article 18 (jan 2023), 23 pages. https://doi.org/10.1145/3519030 ACM Trans. M...

    2023 doi

  49. [2021]

    In Proceedings of the IEEE/CVF international conference on computer vision

    Asymmetric loss for multi-label classification. In Proceedings of the IEEE/CVF international conference on computer vision. 82–91

  50. [2022]

    In Computer Vision – ECCV 2022 , Shai Avidan, Gabriel Brostow, Moustapha Cissé, Giovanni Maria Farinella, and Tal Hassner (Eds.)

    Visual Prompt Tuning. In Computer Vision – ECCV 2022 , Shai Avidan, Gabriel Brostow, Moustapha Cissé, Giovanni Maria Farinella, and Tal Hassner (Eds.). Springer Nature Switzerland, Cham, 709–727

    2022

  51. [2023]

    In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition

    Freeseg: Unified, universal and open-vocabulary image segmentation. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition . 19446–19455

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.