Pith. sign in

REVIEW 3 major objections 6 minor 31 references

LarvSeg: Exploring Image Classification Data For Large Vocabulary Semantic Segmentation via Category-wise Attentive Classifier

T0 review · 3 major / 6 minor · reviewed 2026-08-10 · deepseek-v4-flash

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

Pith's one-line read Image-level category labels can scale semantic segmentation to 21K classes.

desk verdict A solid, useful extension of Detic-style image-level supervision to segmentation, with a genuinely new CA-Classifier and a first 21K-category segmenter; the headline gains are real but the mechanism is under-validated and hyperparameters are tuned on the eval set. read the letter →

arxiv 2501.06862 v1 pith:NIRVAECG submitted 2025-01-12 cs.CV cs.AI

Haojun Yu , Di Dai , Ziwei Zhao , Di He , Han Hu , Liwei Wang This is my paper · ORCID
classification cs.CVcs.AI
keywords LargevocabularysemanticsegmentationCategory-wiseattentiveclassifierImage-levelsupervisionImageNet21KWeaklysupervisedMemorybankprototypes
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 tries to establish that image classification data -- images labeled with category tags rather than pixel masks -- can be used to scale semantic segmentation to vocabularies far beyond the few hundred classes that have mask annotations. It proposes LarvSeg, which trains a segmentation model jointly on pixel-level segmentation data and image-level classification data, and reports that this simple baseline already beats language-guided open-vocabulary segmentation models on categories never seen with masks. The paper further claims that a model trained on only base categories already groups pixels of unseen categories, so a category-wise attentive classifier built from memory-bank prototypes can apply classification supervision to the correct regions. On the reported benchmarks this adds 6.0 mIoU over the baseline on A150 novel categories and 2.1 mIoU on A847 novel categories, and with ImageNet21K it yields, for the first time, a semantic segmentation model for 21K categories. The central thesis is that cheap image-level labels are a practical route to large-vocabulary segmentation.

What carries the argument

The central mechanism is the category-wise attentive classifier (CA-Classifier), a memory-bank prototype module that turns image-level labels into pixel-level supervision. For each novel category it stores representative features taken from pixels with the top-$K$ classification scores across images, computes for each pixel the average cosine similarity to those prototypes, forms a category-wise attention map $A = \mathrm{sigmoid}(\mathrm{norm}(S^{fg}_{\mathrm{mem}} - S^{bg}_{\mathrm{mem}}))$ that strengthens foreground and suppresses background, and applies an auxiliary image-level classification loss to the attentively pooled score map. This is what lets the model supervise the precise regions of categories that never had mask labels.

What would settle it

Evaluate LarvSeg's novel-category mIoU separately for small objects and for categories that usually co-occur with larger background classes, such as 'painting' on a 'wall' or 'mug' on a table. If the CA-Classifier's gain over the simple baseline reverses or disappears on that subset, the assumption that top-$K$ score pixels mark foreground is falsified; the baseline could still work, but the attentive classifier would not be adding correct localization.

Watch

Extended reading notes

Core claim

The paper's central claim is that pixel-level supervision is not necessary for expanding a segmentation model's vocabulary: image-level labels suffice, provided the supervision is aimed at the right pixels. The authors show that a segmenter trained on 171 fully labeled classes, together with weakly labeled classification images, can segment novel categories, and they observe that feature pixels of novel categories are already grouped even when the model never saw those categories. Building on that observation, the category-wise attentive classifier maintains per-category prototypes in a memory bank and uses a foreground-strengthening, background-suppressing attention map so that each novel category's auxiliary image-level loss is concentrated on its actual region. The paper reports that this method outperforms the simple baseline on novel categories, and that training with ImageNet21K yields a 21K-category semantic segmentation model, which the paper states is the first of its kind.

Load-bearing premise

The method assumes that the pixels with the highest classification scores for a category are actually the category's foreground pixels, so the memory-bank prototypes and attention map apply supervision to the right regions rather than to background or frequently co-occurring objects.

Editorial extensions

If this is right

  • Segmentation vocabularies can be expanded with image classification datasets that already exist, avoiding the cost of new pixel-level mask annotation.
  • Training with segmentation masks plus image-level labels outperforms language-guided open-vocabulary models on unseen categories, suggesting image-level supervision is a stronger recognition signal than text alignment.
  • Balanced, object-centric classification data such as ImageNet21K improves novel-category accuracy more than model architecture changes alone, pointing to data quality as the key lever.
  • The category-wise attentive classifier improves segmentation on both multi-label and single-label classification data, and the two kinds of data combine for further gains.
  • A 21K-category segmenter can recognize fine-grained classes such as 'polyphemus moth,' showing the scaling path extends beyond coarse categories.

Reading between the lines

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

  • A reader could take the paper one step further: since the baseline already beats text-supervised models without any language supervision, the practical bottleneck for large-vocabulary segmentation may be cheap category-level labels rather than visual-language alignment; the paper does not state this conclusion explicitly.
  • A testable extension is to shrink the classification dataset per category: if prototypes built from only a few images still lift novel-category mIoU, the approach becomes usable where large classification corpora are unavailable.
  • The evidence that pixel features of unseen categories cluster suggests the same memory-bank mechanism might transfer to panoptic or instance segmentation, where category prototypes could seed region proposals; the paper does not explore that.
  • A stress test on small or heavily co-occurring novel objects would bound the CA-Classifier's added value: a failure there would not refute the baseline but would show the attentive module's localization assumptions.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 6 minor

Summary. The paper proposes LarvSeg, a framework for large-vocabulary semantic segmentation that leverages image classification data with image-level labels to augment a segmentation model trained on pixel-level masks. A simple baseline adds an image-level classification loss on globally pooled features. The paper observes that pixel features of categories outside the pixel-label vocabulary are already clustered, and proposes a category-wise attentive classifier (CA-Classifier) that maintains a memory bank of category prototypes, generates attention maps via foreground/background separation, and applies an auxiliary classification loss to attentively pooled features. Experiments on COCO-Stuff with ADE20K/ADEFull (masks discarded) report that LarvSeg improves over the baseline by 6.0 mIoU on A150 novel categories and 2.1 mIoU on A847 novel categories, and the paper claims the first 21K-category semantic segmentation model trained with ImageNet21K, demonstrated qualitatively.

Significance. The paper introduces a promising and practical direction: using abundant image classification data to scale the vocabulary of segmentation models. The baseline itself is simple, strong, and outperforms prior open-vocabulary methods, which is a meaningful result. The CA-Classifier is an intuitive mechanism with consistent but modest gains. However, the evaluation of the CA-Classifier has important gaps: the core localization premise is not directly validated, and key hyperparameters are tuned on the reported test metric. If these issues are addressed, the paper would be a solid contribution. The code release is a strength and supports reproducibility.

major comments (3)
  1. [Section 3.3, Eqs. (3)-(5)] The central premise that top-K classification-score pixels localize the true foregrounds of novel categories is not validated. The exploratory experiment in Section 3.2 samples a representative pixel from the ground-truth mask (Eq. 2), whereas the actual training procedure updates the memory bank with pixel features that have the highest classification scores, which may not fall inside the object. Since WA150/WA847 are derived from datasets with full masks, a direct audit is feasible: one could compute the precision or IoU of the top-K regions against the original masks. Without such an audit, the gains attributed to the CA-Classifier could stem from the auxiliary loss acting as a regularizer rather than from correct category-wise supervision, and the causal story of the method is unsubstantiated.
  2. [Section 4.5, Table 6] The memory bank size and top-K region area are selected by maximizing the novel-category mIoU on the A150 validation set, which is the same metric used in the headline results (Table 3). This constitutes test-set tuning and can inflate the reported improvement of the CA-Classifier. The paper should either fix these hyperparameters to a priori defaults and report the results, or perform selection on a held-out split and evaluate on another, or report the full sensitivity across a range of values. As written, the 6.0 mIoU improvement on A150 novel categories is not a clean estimate of the method's benefit.
  3. [Section 3.3, Eq. (5)] The auxiliary loss is underspecified. S_cls is not defined as to whether it is the same classification score map used in the baseline or a separate head. The attention map A is defined with a single foreground and background category, but WA150/WA847 are multi-label images containing multiple novel categories; the paper does not explain how fg and bg are determined for each category i, nor how the softmax in Eq. (5) is normalized (over categories or over pixels). The temperature tau is introduced but its value is not listed in Section 4.3. These details are necessary for reproducibility and to assess whether the method is correctly described.
minor comments (6)
  1. [Section 4.3] Equation (5) introduces a temperature tau, but its value is not given in the implementation details; please specify it.
  2. [Section 4.3] The memory bank update rule is not fully described; please state whether features are replaced with moving averages, with what momentum, and whether the top-K selection is performed per image or per batch.
  3. [Table 2] The definitions of A150-C171 and A847-C171 are confusing: they say 'validation set with A150-C171 categories' but the column entries are 2K images and 87/809 classes; please clarify that these are the subsets of the validation set whose categories are not in C171.
  4. [Section 3.1] The description 'we set other categories in the image as ignore labels' is ambiguous for multi-label classification; please state explicitly whether the loss is binary cross-entropy per category with ignore masks.
  5. [References] ADEFull is cited as [29,6], but reference [6] (Cheng et al., 'Per-pixel classification is not all you need') does not introduce ADEFull; please cite the appropriate ADE20K/Scene Parsing dataset papers.
  6. [Abstract / Section 4.6] The claim of being the 'first 21K-category semantic segmentation model' is supported only by qualitative examples; consider softening the claim or providing a quantitative evaluation on a subset with available masks (e.g., ImageNet validation with object masks).

Circularity Check

1 steps flagged · score 3.0 of 10

Mild circularity: memory-bank hyperparameters are selected on the same novel-category mIoU that is then reported as the headline gain; the rest of the method is empirically self-contained.

  1. fitted input called prediction [Section 4.5, Table 6 caption and accompanying ablation text]
    "Table 6: Ablation on memory bank sizes and top-K region areas. We select the hyper-parameters with the best novel categories performance."

    The paper tunes memory bank size M and top-K region area to maximize novel-category mIoU on A150, then uses those settings to report the headline result that 'LarvSeg outperforms the baseline on novel categories by 6.0 mIoU on A150 and 2.1 mIoU on A847.' The reported gain is therefore the maximum over the tested hyperparameter grid of the same metric it is supposed to validate, rather than an out-of-sample prediction from a fixed method. This is a mild form of fitting the evaluation metric: the CA-Classifier's contribution is partly assessed on settings explicitly chosen to make novel-category mIoU as high as possible.

full rationale

The paper is an empirical computer-vision method paper, not a mathematical derivation. The core pipeline — a baseline trained with segmentation plus image-level classification data, augmented by a category-wise attentive classifier that uses a memory bank of top-K scoring pixel features — is a concrete training procedure, and its comparisons against DenseCLIP, ReCo, ZSBaseline, LSeg, and OpenSeg rely on external checkpoints and evaluation protocols rather than on a self-citation chain. No equation defines the reported mIoU in terms of the method's own parameters by construction, and the exploratory experiment in Section 3.2 is a motivation heuristic, not a circular premise. The only circularity-like element is the explicit hyperparameter selection in the ablation study: memory-bank size and top-K are chosen for the best 'novel categories performance,' and the same novel-category mIoU is then used as the headline evidence for the method's effectiveness. This is localized evaluation overfitting rather than derivation circularity: it inflates the reported gain but does not make the method's construction self-referential. The A847 numbers may inherit the same settings without per-dataset tuning, which weakens the concern there. Overall, the central claim has independent empirical content, so the score is 3 rather than higher.

Assumptions & free parameters 5 free parameters · 3 assumptions · 0 invented entities

The central contributions are empirical; no new physical entities are introduced. The main free choices are loss weights and the memory bank hyperparameters, with the latter selected on the evaluation metric. The method assumes that pixel features of unseen categories are compact and that top-K selection in the classification score map corresponds to foreground regions.

free parameters (5)
  • lambda_cls = 0.1
    Set by hand; balances segmentation and image classification losses in Eq. (1); no sensitivity analysis is reported.
  • lambda_aux = 0.1
    Set by hand; weights the auxiliary attention-pooling loss in Eq. (6); no sensitivity analysis is reported.
  • memory_bank_size = 20
    Selected by ablating 10, 20, 40, and 80 and picking the value with the best novel-category mIoU on A150 (Table 6), so the reported gains are partly tuned on the evaluation metric.
  • top_k_region_area = 20
    Selected the same way in Table 6; controls how many high-score pixels update the memory bank.
  • softmax_temperature_tau = Not specified
    Appears in Eq. (5) as the softmax temperature for attentively pooled scores, but its value is not given in the paper.
assumptions (3)
  • domain assumption Pixel features of categories outside the training vocabulary are clustered well enough for retrieval.
    Section 3.2 exploratory experiment (Table 1) claims intra-category compactness for out-of-vocabulary categories is as good as for base categories; this is the premise for using attention maps to locate novel regions.
  • domain assumption Memory bank features updated from top-K classification-score pixels are representative of true category foreground regions.
    The CA-Classifier design in Section 3.3 assumes these prototypes are not corrupted by background or co-occurring categories; this is only indirectly evaluated through the final mIoU.
  • domain assumption ImageNet21K-pretrained ViT-B/16 features provide a useful semantic prior for clustering novel categories.
    The method relies on this pretrained backbone and does not test whether the attention mechanism works with a randomly initialized or weakly pretrained backbone.

how reviews work

0 comments
Cite this review

Pith. "Pith review of LarvSeg: Exploring Image Classification Data For Large Vocabulary Semantic Segmentation via Category-wise Attentive Classifier." pith.science (2026). https://pith.science/paper/NIRVAECG

@misc{pith2026250106862,
  author       = {Pith},
  title        = {Pith review of: LarvSeg: Exploring Image Classification Data For Large Vocabulary Semantic Segmentation via Category-wise Attentive Classifier},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/NIRVAECG}},
  note         = {Machine review of arXiv:2501.06862}
}
read the original abstract

Scaling up the vocabulary of semantic segmentation models is extremely challenging because annotating large-scale mask labels is labour-intensive and time-consuming. Recently, language-guided segmentation models have been proposed to address this challenge. However, their performance drops significantly when applied to out-of-distribution categories. In this paper, we propose a new large vocabulary semantic segmentation framework, called LarvSeg. Different from previous works, LarvSeg leverages image classification data to scale the vocabulary of semantic segmentation models as large-vocabulary classification datasets usually contain balanced categories and are much easier to obtain. However, for classification tasks, the category is image-level, while for segmentation we need to predict the label at pixel level. To address this issue, we first propose a general baseline framework to incorporate image-level supervision into the training process of a pixel-level segmentation model, making the trained network perform semantic segmentation on newly introduced categories in the classification data. We then observe that a model trained on segmentation data can group pixel features of categories beyond the training vocabulary. Inspired by this finding, we design a category-wise attentive classifier to apply supervision to the precise regions of corresponding categories to improve the model performance. Extensive experiments demonstrate that LarvSeg significantly improves the large vocabulary semantic segmentation performance, especially in the categories without mask labels. For the first time, we provide a 21K-category semantic segmentation model with the help of ImageNet21K. The code is available at https://github.com/HaojunYu1998/large_voc_seg.

Figures

Figures reproduced from arXiv: 2501.06862 by the authors.

Figure 1
Figure 1. Illustration of a new paradigm to address large vocabulary semantic [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Visualization of the response maps for pixel grouping. The model is trained [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗
Figure 3
Figure 3. Illustration of LarvSeg framework. The meaning of each icon is listed on [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗
Figures from the paper (2 more)
Figure 4
Figure 4. Figure 4: Visualization of model predictions. The tags show model names and the [PITH_FULL_IMAGE:figures/full_fig_p012_4.png]
Figure 5
Figure 5. Figure 5: Visualization of 21K categories semantic segmentation. [PITH_FULL_IMAGE:figures/full_fig_p012_5.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

31 extracted references · 16 canonical work pages

  1. [1]

    IEEE transactions on pattern analysis and machine intelligence39(12), 2481–2495 (2017)

    Badrinarayanan, V., Kendall, A., Cipolla, R.: Segnet: A deep convolutional encoder- decoder architecture for image segmentation. IEEE transactions on pattern analysis and machine intelligence39(12), 2481–2495 (2017)

    work page 2017

  2. [2]

    In: European conference on computer vision

    Bearman, A., Russakovsky, O., Ferrari, V., Fei-Fei, L.: What’s the point: Semantic segmentation with point supervision. In: European conference on computer vision. pp. 549–565. Springer (2016)

    2016

  3. [3]

    In: Proceedings of the IEEE conference on computer vision and pattern recognition

    Caesar, H., Uijlings, J., Ferrari, V.: Coco-stuff: Thing and stuff classes in context. In: Proceedings of the IEEE conference on computer vision and pattern recognition. pp. 1209–1218 (2018)

    work page 2018

  4. [4]

    In: CVPR (2021)

    Changpinyo, S., Sharma, P., Ding, N., Soricut, R.: Conceptual 12M: Pushing web-scale image-text pre-training to recognize long-tail visual concepts. In: CVPR (2021)

    work page 2021

  5. [5]

    IEEE transactions on pattern analysis and machine intelligence 40(4), 834–848 (2017)

    Chen, L.C., Papandreou, G., Kokkinos, I., Murphy, K., Yuille, A.L.: Deeplab: Semantic image segmentation with deep convolutional nets, atrous convolution, and fully connected crfs. IEEE transactions on pattern analysis and machine intelligence 40(4), 834–848 (2017)

    work page 2017

  6. [6]

    Advances in Neural Information Processing Systems34, 17864–17875 (2021)

    Cheng, B., Schwing, A., Kirillov, A.: Per-pixel classification is not all you need for semantic segmentation. Advances in Neural Information Processing Systems34, 17864–17875 (2021)

    work page 2021

  7. [7]

    In: 2009 IEEE conference on computer vision and pattern recognition

    Deng, J., Dong, W., Socher, R., Li, L.J., Li, K., Fei-Fei, L.: Imagenet: A large-scale hierarchical image database. In: 2009 IEEE conference on computer vision and pattern recognition. pp. 248–255. Ieee (2009)

    2009

  8. [8]

    arXiv preprint arXiv:2208.08984 (2022)

    Ding, Z., Wang, J., Tu, Z.: Open-vocabulary panoptic segmentation with maskclip. arXiv preprint arXiv:2208.08984 (2022)

    arXiv 2022

Show all 31 references
  1. [9]

    arXiv preprint arXiv:2010.11929 (2020)

    Dosovitskiy,A.,Beyer,L.,Kolesnikov,A.,Weissenborn,D.,Zhai,X.,Unterthiner,T., Dehghani, M., Minderer, M., Heigold, G., Gelly, S., et al.: An image is worth 16x16 words: Transformers for image recognition at scale. arXiv preprint arXiv:2010.11929 (2020)

    2020 arXiv

  2. [10]

    arXiv preprint arXiv:2112.12143 (2021)

    Ghiasi, G., Gu, X., Cui, Y., Lin, T.Y.: Open-vocabulary image segmentation. arXiv preprint arXiv:2112.12143 (2021)

    2021 arXiv

  3. [11]

    In: Proceedings of the IEEE conference on computer vision and pattern recognition

    He, K., Zhang, X., Ren, S., Sun, J.: Deep residual learning for image recognition. In: Proceedings of the IEEE conference on computer vision and pattern recognition. pp. 770–778 (2016)

    2016

  4. [12]

    In: International Conference on Machine Learning

    Jia, C., Yang, Y., Xia, Y., Chen, Y.T., Parekh, Z., Pham, H., Le, Q., Sung, Y.H., Li, Z., Duerig, T.: Scaling up visual and vision-language representation learning with noisy text supervision. In: International Conference on Machine Learning. pp. 4904–4916. PMLR (2021)

    2021

  5. [13]

    In: Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV)

    Kirillov, A., Mintun, E., Ravi, N., Mao, H., Rolland, C., Gustafson, L., Xiao, T., Whitehead, S., Berg, A.C., Lo, W.Y., Dollar, P., Girshick, R.: Segment anything. In: Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV). pp. 4015–4026 (October 2023)

    2023

  6. [14]

    arXiv preprint arXiv:2201.03546 (2022)

    Li, B., Weinberger, K.Q., Belongie, S., Koltun, V., Ranftl, R.: Language-driven semantic segmentation. arXiv preprint arXiv:2201.03546 (2022)

    2022 arXiv

  7. [15]

    In: Proceedings of Computer Vision and Pattern Recognition (2023)

    Liang, F., Wu, B., Dai, X., Li, K., Zhao, Y., Zhang, H., Zhang, P., Peter, V., Diana, M.: Open-vocabulary semantic segmentation with mask-adapted clip. In: Proceedings of Computer Vision and Pattern Recognition (2023)

    2023

  8. [16]

    In: European Conference on Computer Vision

    Lin, Y., Li, C., Cao, Y., Zhang, Z., Wang, J., Wang, L., Liu, Z., Hu, H.: A simple approach and benchmark for 21,000-category object detection. In: European Conference on Computer Vision. pp. 1–18. Springer (2022) LarvSeg: Category-wise Attentive Classifier for Image Segmentation 15

    2022

  9. [17]

    In: Proceedings of the IEEE conference on computer vision and pattern recognition

    Long, J., Shelhamer, E., Darrell, T.: Fully convolutional networks for semantic segmentation. In: Proceedings of the IEEE conference on computer vision and pattern recognition. pp. 3431–3440 (2015)

    2015

  10. [18]

    In: ECCV (2020)

    Pont-Tuset, J., Uijlings, J., Changpinyo, S., Soricut, R., Ferrari, V.: Connecting vision and language with localized narratives. In: ECCV (2020)

    2020

  11. [19]

    In: International Conference on Machine Learning

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

    2021

  12. [20]

    arXiv preprint arXiv:2111.02114 (2021)

    Schuhmann, C., Vencu, R., Beaumont, R., Kaczmarczyk, R., Mullis, C., Katta, A., Coombes, T., Jitsev, J., Komatsuzaki, A.: Laion-400m: Open dataset of clip-filtered 400 million image-text pairs. arXiv preprint arXiv:2111.02114 (2021)

    2021 arXiv

  13. [21]

    In: Proceedings of the IEEE conference on computer vision and pattern recognition (2022)

    Shin, G., Xie, W., Albanie, S.: Reco: Retrieve and co-segment for zero-shot transfer. In: Proceedings of the IEEE conference on computer vision and pattern recognition (2022)

    2022

  14. [22]

    In: Proceedings of the IEEE/CVF International Conference on Computer Vision

    Strudel, R., Garcia, R., Laptev, I., Schmid, C.: Segmenter: Transformer for semantic segmentation. In: Proceedings of the IEEE/CVF International Conference on Computer Vision. pp. 7262–7272 (2021)

    2021

  15. [23]

    In: International conference on machine learning

    Tan, M., Le, Q.: Efficientnet: Rethinking model scaling for convolutional neural networks. In: International conference on machine learning. pp. 6105–6114. PMLR (2019)

    2019

  16. [24]

    Communications of the ACM 59(2), 64–73 (2016)

    Thomee, B., Shamma, D.A., Friedland, G., Elizalde, B., Ni, K., Poland, D., Borth, D., Li, L.J.: Yfcc100m: The new data in multimedia research. Communications of the ACM 59(2), 64–73 (2016)

    2016

  17. [25]

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

    Xu, J., De Mello, S., Liu, S., Byeon, W., Breuel, T., Kautz, J., Wang, X.: Groupvit: Semantic segmentation emerges from text supervision. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. pp. 18134– 18144 (2022)

    2022

  18. [26]

    In: Proceedings of Computer Vision and Pattern Recognition (2023)

    Xu, J., Hou, J., Zhang, Y., Feng, R., Wang, Y., Qiao, Y., Xie, W.: Learning open- vocabulary semantic segmentation models from natural language supervision. In: Proceedings of Computer Vision and Pattern Recognition (2023)

    2023

  19. [27]

    arXiv preprint arXiv:2112.14757 (2021)

    Xu, M., Zhang, Z., Wei, F., Lin, Y., Cao, Y., Hu, H., Bai, X.: A simple baseline for zero-shot semantic segmentation with pre-trained vision-language model. arXiv preprint arXiv:2112.14757 (2021)

    2021 arXiv

  20. [28]

    In: European conference on computer vision

    Yuan, Y., Chen, X., Wang, J.: Object-contextual representations for semantic segmentation. In: European conference on computer vision. pp. 173–190. Springer (2020)

    2020

  21. [29]

    In: Proceedings of the IEEE conference on computer vision and pattern recognition

    Zhou, B., Zhao, H., Puig, X., Fidler, S., Barriuso, A., Torralba, A.: Scene parsing through ade20k dataset. In: Proceedings of the IEEE conference on computer vision and pattern recognition. pp. 633–641 (2017)

    2017

  22. [30]

    In: Proceedings of the IEEE conference on computer vision and pattern recognition (2021)

    Zhou, C., Loy, C.C., Dai, B.: Denseclip: Extract free dense labels from clip. In: Proceedings of the IEEE conference on computer vision and pattern recognition (2021)

    2021

  23. [31]

    In: European Conference on Com- puter Vision

    Zhou, X., Girdhar, R., Joulin, A., Krähenbühl, P., Misra, I.: Detecting twenty- thousand classes using image-level supervision. In: European Conference on Com- puter Vision. pp. 350–368. Springer (2022)

    2022

Pith tools

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