Pith. sign in

REVIEW 5 major objections 7 minor 2 cited by

INT: Instance-Specific Negative Mining for Task-Generic Promptable Segmentation

T0 review · 5 major / 7 minor · reviewed 2026-08-09 · deepseek-v4-flash

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

Pith's one-line read A training-free progressive negative mining procedure—erasing candidate objects and multiplying the resulting VLM score drops across iterations—turns one generic prompt into accurate task-specific segmentation without labels, with…

desk verdict A plausible, incremental extension of the authors' own ProMaC work, with consistent gains on six datasets, but the central VLM-drop assumption is under-validated and a few protocol/presentation issues need fixing before acceptance. read the letter →

arxiv 2501.18753 v1 pith:CAZWEKCE submitted 2025-01-30 cs.CV

Jian Hu , Zixu Cheng , Shaogang Gong This is my paper · ORCID
classification cs.CV
keywords task-genericpromptablesegmentationinstance-specificpromptnegativeminingvision-languagemodelstest-timeadaptationtraining-freecamouflagedobjectmedicalimage
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

Task-generic promptable segmentation asks a model to segment every image in a task—say, 'camouflaged animal'—using only that one text prompt, with no per-image labels. This paper proposes INT, a training-free test-time adaptation method that generates candidate instance-specific prompts from patches of the image, then selects among them by erasing the candidate region and measuring how much the vision-language model's output changes. It claims that multiplying these normalized change scores across iterations, a procedure called progressive negative mining, suppresses unreliable candidates and recovers the correct prompt on difficult images. On six datasets spanning camouflaged objects, polyps, and skin lesions, the paper reports state-of-the-art results against prior promptable segmentation methods.

What carries the argument

The load-bearing mechanism is progressive negative mining: for each candidate category, compute $D(y_k^i)$, the maximum drop in the VLM's softmax output when the candidate's predicted region is inpainted; normalize it as $D_{\mathrm{norm}}(y_k^i)=D(y_k^i)/\sum_k D(y_k^i)$; then update the next iteration's score by cumulative multiplication, $D(y_k^{i+1})\leftarrow D(y_k^{i+1})\cdot D_{\mathrm{norm}}(y_k^i)$. The candidate with the largest accumulated score becomes the instance-specific prompt. This multiplication is what lets the method suppress once-lucky wrong categories and reinforce the category that consistently loses unique features under masking.

What would settle it

On a set of images where a large, clearly visible distractor appears next to a small or heavily camouflaged true target, compute $D(y)$ for both candidates under inpainting. If the distractor's score drop exceeds the true target's drop on even a modest fraction of such images, the argmax selection rule (Eq. 5) would pick the wrong prompt; a reader could measure this directly to test whether the assumed monotonicity holds.

Watch

Extended reading notes

Core claim

INT's central claim is that the correct instance-specific prompt is the one whose erasure causes the largest, most stable drop in a VLM's softmax output for that category. The method operationalizes this by hallucinating candidates from overlapping image patches, inpainting each candidate region, and using the normalized difference between pre- and post-inpainting VLM outputs as a selection score; to exploit hard false negative patterns, it multiplies these normalized scores cumulatively over iterations (Eq. 7) so that categories with consistently large differences dominate while sporadic errors decay. The selected prompt drives a semantic mask generator that combines GroundingDINO detections, SAM masks, and CLIP semantic scores, and the resulting soft mask is fed back to focus the next prompt iteration. The paper claims this makes a single task-generic prompt sufficient for accurate segmentation on camouflaged-object and medical datasets, reporting improvements over the prior state of the art on all six evaluated benchmarks.

Load-bearing premise

The assumption that carries the method is that erasing the true target always makes the vision-language model's score for that target drop more than erasing any distractor, and that this ordering remains consistent across iterations.

Editorial extensions

If this is right

  • A single generic prompt ("camouflaged animal", "polyp", "skin lesion") yields per-image, instance-specific segmentation without any manual annotation on six datasets.
  • Erroneous instance-specific prompts produced early in adaptation can be corrected in later iterations, reducing error propagation that earlier VLM-based promptable segmenters suffer from.
  • Because the pipeline is training-free, it can be applied to a new task or dataset at test time with no per-dataset optimization.
  • The method can locate and segment multiple task-related objects within one image, something the paper notes earlier methods could not reliably do.
  • On camouflaged-object benchmarks, INT reports results above both point- and scribble-supervised weakly supervised methods, despite using no training labels at all.

Reading between the lines

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

  • A natural extension would use the accumulated negative-mining score as a per-image confidence signal, flagging low-confidence cases for human review; the paper does not report such a use.
  • The method's behavior likely depends on the inpainting model's quality: if inpainting fails to plausibly fill a region, the measured output drop may reflect artifacts rather than the absence of the object, and testing with weaker inpainting would clarify how much of the gain comes from the mining rule versus the erasure tool.
  • Whether progressive multiplication helps or hurts probably hinges on the first iteration: a wrong candidate that scores large at $i=1$ gets amplified, so perturbing or randomizing the initial candidate ranking could reveal how robust the correction is.
  • If the assumption generalizes beyond natural and medical images, the same machinery could be applied to other VLM-driven dense prediction tasks such as referring expression segmentation, though the paper does not test this.
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

5 major / 7 minor

Summary. The paper introduces INT, a training-free test-time adaptation method for task-generic promptable image segmentation, requiring only a single task-generic prompt such as 'camouflaged animal' or 'polyp'. The method generates candidate instance-specific prompts by processing multi-scale image patches with a VLM, selects the prompt whose category shows the largest VLM output drop after inpainting candidate-indicated regions, and progressively multiplies normalized per-iteration differences to suppress unstable wrong categories. The selected prompt is then used with GroundingDINO, SAM, and CLIP to form a semantic mask, which is fed back as an inpainting guide and image weighting for the next iteration. Experiments on CHAMELEON, CAMO, COD10K, CVC-ColonDB, Kvasir, and ISIC report consistent improvements over ProMaC and other task-generic baselines, with ablations on CHAMELEON attributing gains to each module.

Significance. If the reported gains are reliable, INT provides a practical and conceptually simple extension of promptable segmentation by using VLM response changes as a no-label quality signal for instance-specific prompts. The method is training-free, builds on publicly available components (LLaVA, SAM, GroundingDINO, CLIP, Stable Diffusion), and demonstrates consistent numerical improvements across six datasets and three task types. The paper also makes an explicit falsifiable claim: that progressive multiplication of normalized VLM output differences amplifies stable correct-category cues over sporadic wrong-category cues. That claim is central to the method and is currently supported only by an illustrative figure using ground-truth masks, not by quantitative validation on predicted masks, so the significance of the result depends on additional evidence rather than on the current manuscript alone.

major comments (5)
  1. [§4.1 vs Table 3(a)] Section 4.1 states that all tasks undergo four adaptation epochs, except polyp segmentation, which uses six. However, the CHAMELEON results in Table 1 (M=0.039, Fβ=0.801, Eφ=0.906, Sα=0.842) are exactly the I=5 row of Table 3(a), not the I=4 row (M=0.045, Fβ=0.792, Eφ=0.903, Sα=0.829). Please clarify which iteration count produced each main result and reconcile the text with the ablation table.
  2. [§4.1] The text says 'Our results are the average of three trials,' but no standard deviations, confidence intervals, or significance tests are reported in Tables 1–3. Several margins over ProMaC are small (e.g., CAMO M: 0.086 vs 0.090; ISIC Sα: 0.708 vs 0.703), so without variance estimates the reader cannot tell whether the improvements are beyond run-to-run noise. Please provide per-trial results or error bars for at least the main tables.
  3. [§3.1, Eqs. (3)–(7)] The central assumption that the magnitude of the VLM output drop after inpainting is a monotone indicator of prompt correctness is not validated on predicted masks. Figure 1(b) is computed with ground-truth masks, which are unavailable at test time, while Eq. (4) uses the predicted mask from the previous iteration. If a predicted mask removes a salient distractor but misses the true object, the distractor category can show a large drop and the correct category a small drop, so Eq. (5) can select the wrong prompt and Eq. (7) can amplify it over iterations. Please provide a direct evaluation of this failure mode, e.g., prompt-selection accuracy against ground-truth categories per iteration, or the fraction of samples where iterative multiplication reinforces an incorrect category.
  4. [§3.1, Eqs. (6)–(7)] The mathematical indexing in the progressive negative mining step is unclear. Equation (6) normalizes D(y_i^k) over the K patches, but the candidate set is described as categories; Eq. (7) then writes D(y_{i+1}^k) = D(y_{i+1}^k) · Dnorm(y_i^k), using the same symbol D for both the raw and updated scores. Please clarify how candidate categories are matched across patches and iterations, and rewrite the update so that the raw difference, the normalized weight, and the cumulative product are distinct variables.
  5. [§3.2, Eq. (13)] The final mask is selected as the iteration mask closest to the average of all iteration masks. This rule can discard a genuinely better late-iteration mask that is an outlier relative to the average, and no sensitivity analysis is provided for this choice. Please report how the main results change if the last-iteration mask, or a mean mask, is used instead.
minor comments (7)
  1. [Abstract and §1] There are several typos: 'fliters' should be 'filters' in the Abstract, 'Specfically' should be 'Specifically' in Section 1, 'distintive' should be 'distinctive', and 'propogation' should be 'propagation'.
  2. [Table 3(b)] The label 'Havel' should read 'Halve' to match the text describing horizontal or vertical splitting.
  3. [§2] 'A V-SAM' should be 'AV-SAM' for the audio-visual SAM reference.
  4. [Table 2] The venue label 'NeruIPS' in the LLaVA1.5+SAM row should be 'NeurIPS'.
  5. [§4.1] The sentence 'Our study evaluates the ProMaC model’s effectiveness' appears to refer to INT rather than ProMaC; please correct the naming.
  6. [§3.1, Eq. (2)] The caption 'C k' in Eq. (2) is not defined, and the output structure of the VLM calls (what exactly 'Bk' contains, how many boxes are returned) is left unspecified; please define these quantities.
  7. [§3.2, Eq. (12)] The image weighting hyperparameter w is set to 0.3 without a sensitivity study; even a brief table over a small range of w would help establish robustness.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: progressive negative mining is an independent, training-free derivation evaluated against external ground-truth masks.

full rationale

INT's central claim is that cumulative multiplication of normalized VLM output differences before and after inpainting selects better instance-specific prompts. Equation 3 defines the difference, Eq. 6 normalizes it, Eq. 7 accumulates it, and Eq. 5 selects the argmax; no test-set metric enters these equations. The evaluation in Tables 1 and 2 compares against external datasets (CHAMELEON, CAMO, COD10K, CVC-ColonDB, Kvasir, ISIC) and multiple baselines, so the result is not fitted into the method. The method reuses the authors' prior GenSAM Spatial CLIP (Eq. 9) and ProMaC-style patch candidate generation (Eq. 2), but these are modular components, not the load-bearing proof of negative mining; the central novelty is independently evaluated. The assumption that larger VLM-output drops indicate correct candidates (Fig. 1) is an empirical hypothesis rather than a definitional equivalence, and its failure would degrade the method instead of making the derivation circular. No self-citation chain forces the reported outcome.

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

The central claim rests on several empirical assumptions about VLM behavior and pre-trained components. These are testable design assumptions rather than hidden curve-fitting, but they are not derived or separately validated in the paper.

free parameters (3)
  • Mask weighting factor w = 0.3
    Set by hand in Eq. 12 to blend the current mask into the next iteration's input image; no sensitivity analysis is reported.
  • Number of test-time adaptation iterations I = 4 in the setup text; 5 used for CHAMELEON results; 6 for polyps
    Section 4.1 says four epochs generally and six for polyps, but Table 3(a) sweeps I on CHAMELEON and the reported CHAMELEON result matches I=5, suggesting selection on the test set.
  • Patch splitting scales = Original + Halve + Quarters
    Chosen based on Table 3(b) as the best combination on CHAMELEON; other combinations were tested and discarded.
assumptions (5)
  • domain assumption VLM output change after masking a region is a reliable proxy for prompt correctness, with larger changes indicating more accurate candidates.
    Introduced in Section 3.1 and Figure 1; used to define the selection score in Eqs. 3-7. It is plausible but not independently validated.
  • domain assumption Patch-based VLM candidate generation produces a candidate list that contains the correct instance-specific prompt at least once.
    Stated in Section 3.1 as the need to identify as many candidates as possible so that the correct prompt is not overlooked. If the correct object is never named, selection and negative mining cannot recover it.
  • domain assumption CLIP semantic similarity between a masked image and the instance prompt is a valid quality signal for ranking masks.
    Used in Eqs. 10-11 to weight and combine SAM masks; no calibration or failure analysis is provided.
  • domain assumption Stable Diffusion inpainting removes the target object and preserves the surrounding background well enough not to distort VLM scoring.
    Required in Eq. 4 to produce X'_k for the before/after comparison; failures in inpainting would corrupt the selection scores.
  • domain assumption GroundingDINO and SAM provide adequate base object localization and mask proposals for all six datasets.
    Used in Eqs. 8-9 as fixed components; no analysis of their failure modes in this pipeline is given.

how reviews work

0 comments
Cite this review

Pith. "Pith review of INT: Instance-Specific Negative Mining for Task-Generic Promptable Segmentation." pith.science (2026). https://pith.science/paper/CAZWEKCE

@misc{pith2026250118753,
  author       = {Pith},
  title        = {Pith review of: INT: Instance-Specific Negative Mining for Task-Generic Promptable Segmentation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/CAZWEKCE}},
  note         = {Machine review of arXiv:2501.18753}
}
read the original abstract

Task-generic promptable image segmentation aims to achieve segmentation of diverse samples under a single task description by utilizing only one task-generic prompt. Current methods leverage the generalization capabilities of Vision-Language Models (VLMs) to infer instance-specific prompts from these task-generic prompts in order to guide the segmentation process. However, when VLMs struggle to generalise to some image instances, predicting instance-specific prompts becomes poor. To solve this problem, we introduce \textbf{I}nstance-specific \textbf{N}egative Mining for \textbf{T}ask-Generic Promptable Segmentation (\textbf{INT}). The key idea of INT is to adaptively reduce the influence of irrelevant (negative) prior knowledge whilst to increase the use the most plausible prior knowledge, selected by negative mining with higher contrast, in order to optimise instance-specific prompts generation. Specifically, INT consists of two components: (1) instance-specific prompt generation, which progressively fliters out incorrect information in prompt generation; (2) semantic mask generation, which ensures each image instance segmentation matches correctly the semantics of the instance-specific prompts. INT is validated on six datasets, including camouflaged objects and medical images, demonstrating its effectiveness, robustness and scalability.

Figures

Figures reproduced from arXiv: 2501.18753 by the authors.

Figure 1
Figure 1. (a) Motivation of INT. When task-related objects in the [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. INT consists of two main components: instance-specific prompt generation and semantic mask generation. Initially, the former [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Visualization of various segmentation methods among various segmentation tasks. [PITH_FULL_IMAGE:figures/full_fig_p007_3.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 2 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. ArgusCogito: Chain-of-Thought for Cross-Modal Synergy and Omnidirectional Reasoning in Camouflaged Object Segmentation

    cs.CV 2025-08 conditional novelty 6.0 of 10

    ArgusCogito combines a vision-language model's reasoning with depth cues and iterative point prompts to segment camouflaged objects without any task-specific training.

  2. Seg-R1: Segmentation Can Be Surprisingly Simple with Reinforcement Learning

    cs.CV 2025-06 conditional novelty 5.0 of 10

    Reinforcement learning can teach an LMM to prompt SAM2 for segmentation, achieving competitive camouflaged and salient object detection and zero-shot referring segmentation.

Reference graph

Works this paper leans on

38 extracted references · 22 canonical work pages · cited by 2 Pith papers

  1. [1]

    Flamingo: a visual language model for few-shot learning

    [Alayrac et al., 2022] Jean-Baptiste Alayrac, Jeff Donahue, Pauline Luc, Antoine Miech, Iain Barr, Yana Hasson, Karel Lenc, Arthur Mensch, Katherine Millican, Malcolm Reynolds, et al. Flamingo: a visual language model for few-shot learning. Advances in Neural Information Pro- cessing Systems, 35:23716–23736,

    work page 2022

  2. [5]

    Multi-weight partial domain adaptation

    [Hu et al., 2019] Jian Hu, Hongya Tuo, Chao Wang, Lingfeng Qiao, Haowen Zhong, and Zhongliang Jing. Multi-weight partial domain adaptation. InBMVC, page 5,

    work page 2019

  3. [6]

    Discriminative partial domain ad- versarial network

    [Hu et al., 2020a] Jian Hu, Hongya Tuo, Chao Wang, Lingfeng Qiao, Haowen Zhong, Junchi Yan, Zhongliang Jing, and Henry Leung. Discriminative partial domain ad- versarial network. In Computer Vision–ECCV 2020: 16th European Conference, Glasgow, UK, August 23–28, 2020, Proceedings, Part XXVII 16 , pages 632–648. Springer,

    work page 2020

  4. [9]

    Kvasir-seg: A segmented polyp dataset

    [Jha et al., 2020] Debesh Jha, Pia H Smedsrud, Michael A Riegler, P˚al Halvorsen, Thomas de Lange, Dag Johansen, and H˚avard D Johansen. Kvasir-seg: A segmented polyp dataset. In MultiMedia Modeling: 26th International Con- ference, MMM 2020, Daejeon, South Korea, January 5–8, 2020, Proceedings, Part II 26 , pages 451–462. Springer,

    work page 2020

  5. [10]

    Segment anything is not always perfect: An investigation of sam on different real-world applications

    [Ji et al., 2023] Wei Ji, Jingjing Li, Qi Bi, Wenbo Li, and Li Cheng. Segment anything is not always perfect: An investigation of sam on different real-world applications. arXiv preprint arXiv:2304.05750,

    arXiv 2023

  6. [11]

    Visual prompt tuning

    [Jia et al., 2022] Menglin Jia, Luming Tang, Bor-Chun Chen, Claire Cardie, Serge Belongie, Bharath Hariharan, and Ser-Nam Lim. Visual prompt tuning. In European Conference on Computer Vision, pages 709–727. Springer,

    work page 2022

  7. [12]

    Segment anything

    [Kirillov et al., 2023] Alexander Kirillov, Eric Mintun, Nikhila Ravi, Hanzi Mao, Chloe Rolland, Laura Gustafson, Tete Xiao, Spencer Whitehead, Alexander C Berg, Wan-Yen Lo, et al. Segment anything. arXiv preprint arXiv:2304.02643,

    arXiv 2023

  8. [13]

    Imagenet classification with deep convolutional neural networks

    [Krizhevsky et al., 2012] Alex Krizhevsky, Ilya Sutskever, and Geoffrey E Hinton. Imagenet classification with deep convolutional neural networks. Advances in neural infor- mation processing systems, 25,

    2012

Show all 38 references
  1. [15]

    Generated knowledge prompting for commonsense reasoning

    [Liu et al., 2021] Jiacheng Liu, Alisa Liu, Ximing Lu, Sean Welleck, Peter West, Ronan Le Bras, Yejin Choi, and Hannaneh Hajishirzi. Generated knowledge prompting for commonsense reasoning. arXiv preprint arXiv:2110.08387,

    2021 arXiv

  2. [16]

    Visual instruction tuning

    [Liu et al., 2023b] Haotian Liu, Chunyuan Li, Qingyang Wu, and Yong Jae Lee. Visual instruction tuning. arXiv preprint arXiv:2304.08485,

    arXiv

  3. [17]

    Grounding dino: Marrying dino with grounded pre-training for open-set object detection

    [Liu et al., 2023c] Shilong Liu, Zhaoyang Zeng, Tianhe Ren, Feng Li, Hao Zhang, Jie Yang, Chunyuan Li, Jianwei Yang, Hang Su, et al. Grounding dino: Marrying dino with grounded pre-training for open-set object detection. arXiv preprint arXiv:2303.05499,

    arXiv

  4. [18]

    Graphprompt: Unifying pre-training and downstream tasks for graph neural networks

    [Liu et al., 2023d] Zemin Liu, Xingtong Yu, Yuan Fang, and Xinming Zhang. Graphprompt: Unifying pre-training and downstream tasks for graph neural networks. In Proceed- ings of the ACM Web Conference 2023 , pages 417–428,

    2023

  5. [19]

    Understanding and mitigating overfitting in prompt tuning for vision-language models

    [Ma et al., 2023] Chengcheng Ma, Yang Liu, Jiankang Deng, Lingxi Xie, Weiming Dong, and Changsheng Xu. Understanding and mitigating overfitting in prompt tuning for vision-language models. IEEE Transactions on Cir- cuits and Systems for Video Technology,

    2023

  6. [20]

    How to evaluate foreground maps? In Proceedings of the IEEE conference on computer vision and pattern recognition, pages 248–255,

    [Margolin et al., 2014] Ran Margolin, Lihi Zelnik-Manor, and Ayellet Tal. How to evaluate foreground maps? In Proceedings of the IEEE conference on computer vision and pattern recognition, pages 248–255,

    2014

  7. [22]

    [OpenAI, 2024b] OpenAI

    Accessed: 2024-05-20. [OpenAI, 2024b] OpenAI. Hello gpt-4o

    2024

  8. [23]

    [Radford et al., 2021] Alec Radford, Jong Wook Kim, Chris Hallacy, Aditya Ramesh, Gabriel Goh, Sandhini Agar- wal, Girish Sastry, Amanda Askell, Pamela Mishkin, Jack Clark, et al

    Accessed: 2024-05-20. [Radford et al., 2021] Alec Radford, Jong Wook Kim, Chris Hallacy, Aditya Ramesh, Gabriel Goh, Sandhini Agar- wal, Girish Sastry, Amanda Askell, Pamela Mishkin, Jack Clark, et al. Learning transferable visual models from nat- ural language supervision. In...

    2024

  9. [25]

    Animal camouflage analysis: Chameleon database

    [Skurowski et al., 2018] Przemysław Skurowski, Hassan Abdulameer, J Błaszczyk, Tomasz Depta, Adam Kornacki, and P Kozieł. Animal camouflage analysis: Chameleon database. Unpublished manuscript, 2(6):7,

    2018

  10. [28]

    Finetuned lan- guage models are zero-shot learners

    [Wei et al., 2021] Jason Wei, Maarten Bosma, Vincent Y Zhao, Kelvin Guu, Adams Wei Yu, Brian Lester, Nan Du, Andrew M Dai, and Quoc V Le. Finetuned lan- guage models are zero-shot learners. arXiv preprint arXiv:2109.01652,

    2021 arXiv

  11. [29]

    Chain-of-thought prompting elicits reasoning in large language models

    [Wei et al., 2022] Jason Wei, Xuezhi Wang, Dale Schuur- mans, Maarten Bosma, Fei Xia, Ed Chi, Quoc V Le, Denny Zhou, et al. Chain-of-thought prompting elicits reasoning in large language models. Advances in Neural Information Processing Systems, 35:24824–24837,

    2022

  12. [30]

    Class-aware visual prompt tuning for vision-language pre- trained model

    [Xing et al., 2022] Yinghui Xing, Qirui Wu, De Cheng, Shizhou Zhang, Guoqiang Liang, and Yanning Zhang. Class-aware visual prompt tuning for vision-language pre- trained model. arXiv preprint arXiv:2208.08340,

    2022 arXiv

  13. [31]

    Structure-consistent weakly supervised salient object detection with local saliency coherence

    [Yu et al., 2021] Siyue Yu, Bingfeng Zhang, Jimin Xiao, and Eng Gee Lim. Structure-consistent weakly supervised salient object detection with local saliency coherence. In Proceedings of the AAAI conference on artificial intelli- gence, volume 35, pages 3234–3242,

    2021

  14. [32]

    Unified vi- sion and language prompt learning

    [Zang et al., 2022] Yuhang Zang, Wei Li, Kaiyang Zhou, Chen Huang, and Chen Change Loy. Unified vi- sion and language prompt learning. arXiv preprint arXiv:2210.07225,

    2022 arXiv

  15. [33]

    Weakly-supervised salient object detection via scribble annotations

    [Zhang et al., 2020] Jing Zhang, Xin Yu, Aixuan Li, Peipei Song, Bowen Liu, and Yuchao Dai. Weakly-supervised salient object detection via scribble annotations. In Pro- ceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 12546–12555,

    2020

  16. [34]

    Domain adaptive yolo for one-stage cross-domain detection

    [Zhang et al., 2021] Shizhao Zhang, Hongya Tuo, Jian Hu, and Zhongliang Jing. Domain adaptive yolo for one-stage cross-domain detection. In Asian conference on machine learning, pages 785–797. PMLR,

    2021

  17. [35]

    Multimodal chain-of-thought reasoning in language models

    [Zhang et al., 2023] Zhuosheng Zhang, Aston Zhang, Mu Li, Hai Zhao, George Karypis, and Alex Smola. Multimodal chain-of-thought reasoning in language models. arXiv preprint arXiv:2302.00923,

    2023 arXiv

  18. [36]

    Learning to prompt for vision-language models

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

    2022

  19. [37]

    Zegclip: Towards adapting clip for zero-shot semantic segmentation

    [Zhou et al., 2023] Ziqin Zhou, Yinjie Lei, Bowen Zhang, Lingqiao Liu, and Yifan Liu. Zegclip: Towards adapting clip for zero-shot semantic segmentation. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pat- tern Recognition, pages 11175–11185,

    2023

  20. [38]

    Segment everything everywhere all at once

    [Zou et al., 2023b] Xueyan Zou, Jianwei Yang, Hao Zhang, Feng Li, Linjie Li, Jianfeng Gao, and Yong Jae Lee. Segment everything everywhere all at once. arXiv:2304.06718, 2023

    2023 arXiv

  21. [2012]

    Anabranch network for camouflaged object segmentation

    [Le et al., 2019] Trung-Nghia Le, Tam V Nguyen, Zhongliang Nie, Minh-Triet Tran, and Akihiro Sug- imoto. Anabranch network for camouflaged object segmentation. Computer vision and image understanding, 184:45–56,

    2019

  22. [2014]

    Av- sam: Segment anything model meets audio-visual local- ization and segmentation

    [Mo and Tian, 2023] Shentong Mo and Yapeng Tian. Av- sam: Segment anything model meets audio-visual local- ization and segmentation. arXiv:2305.01836,

    2023 arXiv

  23. [2015]

    Self-consistency improves chain of thought reasoning in language models

    [Wang et al., 2022] Xuezhi Wang, Jason Wei, Dale Schu- urmans, Quoc Le, Ed Chi, Sharan Narang, Aakanksha Chowdhery, and Denny Zhou. Self-consistency improves chain of thought reasoning in language models. arXiv preprint arXiv:2203.11171,

    2022 arXiv

  24. [2018]

    Automated polyp detection in colonoscopy videos using shape and context informa- tion

    [Tajbakhsh et al., 2015] Nima Tajbakhsh, Suryakanth R Gu- rudu, and Jianming Liang. Automated polyp detection in colonoscopy videos using shape and context informa- tion. IEEE transactions on medical imaging , 35(2):630– 644,

    2015

  25. [2019]

    Structure-measure: A new way to evaluate foreground maps

    [Fan et al., 2017] Deng-Ping Fan, Ming-Ming Cheng, Yun Liu, Tao Li, and Ali Borji. Structure-measure: A new way to evaluate foreground maps. In Proceedings of the IEEE international conference on computer vision , pages 4548–4557,

    2017

  26. [2020]

    Learning unbiased transferability for domain adaptation by uncertainty mod- eling

    [Hu et al., 2022] Jian Hu, Haowen Zhong, Fei Yang, Shao- gang Gong, Guile Wu, and Junchi Yan. Learning unbiased transferability for domain adaptation by uncertainty mod- eling. In European Conference on Computer Vision, pages 223–241. Springer,

    2022

  27. [2021]

    Zero-shot text-to-image generation

    [Ramesh et al., 2021] Aditya Ramesh, Mikhail Pavlov, Gabriel Goh, Scott Gray, Chelsea V oss, Alec Radford, Mark Chen, and Ilya Sutskever. Zero-shot text-to-image generation. In International Conference on Machine Learning, pages 8821–8831. PMLR,

    2021

  28. [2022]

    Skin lesion analysis to- ward melanoma detection 2018: A challenge hosted by the international skin imaging collaboration (isic)

    [Codella et al., 2019] Noel Codella, Veronica Rotemberg, Philipp Tschandl, M Emre Celebi, Stephen Dusza, David Gutman, Brian Helba, Aadi Kalloo, Konstantinos Liopy- ris, Michael Marchetti, et al. Skin lesion analysis to- ward melanoma detection 2018: A challenge hosted by the ...

    2019 arXiv

  29. [2023]

    Weakly-supervised concealed object segmen- tation with sam-based pseudo labeling and multi-scale fea- ture grouping

    [He et al., 2023b] Chunming He, Kai Li, Yachao Zhang, Guoxia Xu, Longxiang Tang, Yulun Zhang, Zhenhua Guo, and Xiu Li. Weakly-supervised concealed object segmen- tation with sam-based pseudo labeling and multi-scale fea- ture grouping. arXiv preprint arXiv:2305.11003,

    arXiv

  30. [2024]

    Leveraging hallucinations to reduce manual prompt dependency in promptable segmentation

    [Hu et al., 2024b] Jian Hu, Jiayi Lin, Junchi Yan, and Shao- gang Gong. Leveraging hallucinations to reduce manual prompt dependency in promptable segmentation. arXiv preprint arXiv:2408.15205,

Pith tools

Reviewed August 9, 2026 · model on record in the stance chip above.