Pith. sign in

REVIEW 3 major objections 5 minor 102 references

Inspiring the Next Generation of Segment Anything Models: Comprehensively Evaluate SAM and SAM 2 with Diverse Prompts Towards Context-Dependent Concepts under Different Scenes

T0 review · 3 major / 5 minor · reviewed 2026-08-12 · deepseek-v4-flash

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

Pith's one-line read This evaluation maps SAM and SAM 2's performance boundaries on 11 context-dependent segmentation concepts, finding box prompts generally best and SAM 2's edge confined to video and 3D.

desk verdict Broad, useful SAM/SAM2 evaluation map, but headline performance boundaries are GT-conditioned upper envelopes and two summary claims contradict the paper's own tables. read the letter →

arxiv 2412.01240 v3 pith:H72YTHPZ submitted 2024-12-02 cs.CV

Xiaoqi Zhao , Youwei Pang , Shijie Chang , Yuan Zhao , Lihe Zhang , Chenyang Yu , Hanqi Liu , Jiaming Zuo
show 5 more authors
Jinsong Ouyang Weisi Lin Georges El Fakhri Huchuan Lu Xiaofeng Liu
This is my paper · ORCID
classification cs.CV
keywords SegmentAnythingModelSAM2context-dependentconceptspromptrobustnessin-contextlearningsegmentationevaluationbidirectionalinferencebenchmark
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

This paper tries to establish where the Segment Anything Model (SAM) and its video-capable successor SAM 2 actually stand when the things to segment are context-dependent: visual saliency, camouflage, shadows, transparent objects, medical lesions, and industrial defects. Across 33 datasets, 11 concepts, and image, video, and 3D data, the authors build a unified prompting framework and produce a performance map. Their central findings are that box prompts are the most reliable prompt type, that SAM 2 does not uniformly beat SAM and is weaker on everything-mode and point-prompt tasks, and that SAM 2's temporal memory becomes a decisive advantage in video and 3D once multiple mask prompts and bidirectional inference are used. The value is a reference baseline for deploying SAMs on real-world context-dependent tasks and guidance for the next generation of the model.

What carries the argument

The load-bearing mechanism is the unified evaluation framework itself: six prompting settings (everything, mask, box, point, in-context learning, and prompt robustness) applied to SAM and SAM 2 through three prompt-generation strategies. The prediction-based propagated prompt turns the previous frame's predicted mask into the next frame's prompt, letting an image-trained SAM behave as a video segmenter; the bidirectional inference strategy picks the slice with the largest ground-truth foreground mask as an anchor and propagates masks in both directions through a 3D volume; and the in-context learning mode feeds SAM 2 the first 20 training images and masks of a concept as exemplars. The everything-mode outputs are filtered by an overlap filtering strategy that keeps only predicted entities whose overlap with the ground truth exceeds 90%. These mechanisms together convert frozen, off-the-shelf SAMs into measurable systems across heterogeneous data without task-specific fine-tuning.

What would settle it

Rerun the headline settings on a held-out subset of these datasets with no oracle: prompt points and boxes from an off-the-shelf detector or saliency model, no overlap filtering, and no ground-truth-based anchor selection; if SAM 2's everything-mode and point-mode deficits and its 3D advantage shrink or reverse, the central conclusions are artifacts of ground-truth-conditioned prompting. Independently, re-extract the specialist baselines in Table 15a from the original papers, since identical scores across all four modalities would indicate transcription error.

Watch

Extended reading notes

Core claim

The paper's central claim is that SAM and SAM 2 have clear, measurable performance boundaries on context-dependent concepts, and that those boundaries depend more on prompt type, data modality, and target material than on model scale. In static images, box prompts dominate on 10 of the 11 concepts and rescue performance on hard cases like camouflage and medical lesions. In everything mode and point-click mode, SAM 2 often performs worse than SAM, which the authors attribute to temporal memory and dynamic prompt attention adding unfavorable bias when no temporal context is available. In video and 3D, the picture flips: SAM 2 with prediction-based propagated prompts, multi-frame mask prompts, and a bidirectional inference strategy for volumetric data surpasses task-specific medical segmentors on brain tumor and multiple sclerosis lesion segmentation. The paper also claims SAM 2 shows real but incomplete in-context learning ability, and that both models are highly sensitive to imperfect prompts, degrading sharply under small perturbations.

Load-bearing premise

The headline numbers assume that ground-truth-derived prompts and ground-truth-filtered outputs are the right way to measure SAMs' context-dependent segmentation ability, and if those oracle conditions are removed, the reported performance boundaries may not describe realistic zero-shot or interactive use.

Editorial extensions

If this is right

  • Box prompts should be the default interface when deploying SAMs on context-dependent image tasks; point and everything modes should be treated as unreliable outside well-separated objects.
  • SAM 2's temporal memory pays off only when there is temporal structure; for single-frame inference, the original SAM remains competitive or better, so upgrading to SAM 2 is not automatically an improvement.
  • In video and 3D medical segmentation, SAM 2 with mask prompts and multi-frame propagation can serve as a strong zero-shot baseline, competitive with or better than task-specific networks.
  • Evaluation practices should include prompt-robustness testing, because small perturbations to boxes, points, or masks move scores by percentages that change qualitative conclusions.
  • The in-context learning results suggest that a few exemplars can steer SAM 2 toward new concepts, though not yet to the level of existing unified models.

Reading between the lines

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

  • If the oracle-conditioned prompts inflate performance, then the real-world gap between SAMs and specialized models on camouflage, shadows, and medical lesions is larger than the tables suggest; a fully automatic prompt-setting evaluation would test this directly.
  • The finding that SAM 2 underperforms SAM on everything mode in static images hints that the memory module or dynamic prompt attention introduces a temporal prior that is harmful without temporal input; inspecting memory-bank activity on single frames could localize the cause.
  • The box-prompt dominance across 10 of 11 concepts suggests SAMs act essentially as strong object proposal matchers: given spatial bounds they segment well, but their ability to discover context-dependent regions on their own is limited, so adding a saliency or anomaly proposal module as an automatic prompt generator could be a practical route toward a better next-generation model.
  • The 3D success with bidirectional multi-frame mask prompts implies that medical volumes can be treated as videos without architectural changes; testing on CT and MRI modalities beyond the ones reported would establish whether that advantage generalizes.
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 / 5 minor

Summary. This manuscript presents a large-scale empirical evaluation of SAM and SAM 2 on eleven context-dependent (CD) segmentation concepts across 33 datasets, covering 2D images, videos, and 3D medical volumes in natural, medical, and industrial scenes. The authors propose a unified prompting framework that includes everything, box, point, mask, in-context learning, and prompt-robustness modes, and they distill the results into design insights for a future SAM 3. The headline conclusions are that box prompts are generally the most advantageous prompt type, that SAM 2 is not uniformly better than SAM and is worse on everything and point modes, that SAM 2 shows in-context-learning potential, and that bidirectional inference with multi-frame mask prompts enables SAM 2 to surpass specialized 3D medical segmentation models.

Significance. The benchmark is timely and potentially useful: it is the most comprehensive CD-concept evaluation of SAM and SAM 2 that I am aware of, it covers underexplored medical and industrial modalities, and it ships a unified evaluation toolkit and code repository, which supports reproducibility. The robustness analysis in Section 3.8 and the in-context-learning exploration in Section 3.4.2 go beyond earlier SAM evaluations and provide practically relevant information for deployment. If the headline findings were established by the reported data, they would give the community a reference map for prompt selection on CD concepts and a concrete evidence base for SAM 3 design; that is a meaningful contribution. However, as detailed below, several load-bearing conclusions outrun the ground-truth-conditioned protocols that produced the numbers, and one stated conclusion is contradicted by the paper's own tables.

major comments (3)
  1. [§3.4.1 and §3.9, Items I, II, V] The headline performance conclusions are derived under ground-truth-conditioned protocols rather than zero-shot or realistic interactive conditions. In automatic everything mode, only predicted entities whose overlap with the GT is greater than 90% are retained and merged (Sec. 3.4.1); in point mode, corrective clicks are placed at the largest GT-error region until IoU reaches 0.9 or six clicks are used (Sec. 3.4.1); and in 3D, the anchor slice is selected as the one with the largest GT foreground (Sec. 3.4.4). Consequently, claims such as "box prompts are generally the most advantageous" and "SAM 2 performs worse on everything and point prompts" describe behavior under oracle feedback, not the open-world or zero-shot behavior implied by the abstract and Section 1. The everything-mode comparison may also be confounded by the OFS granularity test: SAM 2's more fragmented automatic masks would be discarded by the >90% per-entity overlap rule even when their union is correct. Section 5's limitations do not disclose this GT-conditioning. Please either re-run automatic and point modes without GT-based filtering/correction (or with a realistic interaction model), re-evaluate the 3D anchor choice without GT, or restrict the conclusions to the oracle-conditioned setting and revise the abstract and Section 3.9 accordingly.
  2. [§3.5 and Tables 4, 9] The statement that "SAM 2 (point) and SAM 2 (everything) are consistently weaker than their corresponding SAM variants" is contradicted by the paper's own tables. For point mode, Table 4 reports COD10K F_beta of 0.864 for SAM 2 versus 0.823 for SAM, and Table 9 reports SAM 2 with higher Dice on COVID-19 (0.687 vs 0.352), BUSI (0.783 vs 0.694), ISIC-2018 (0.641 vs 0.504), and Polyp-Five (0.862 vs 0.641). For everything mode, the comparison is also uneven across tasks and datasets. The claim in Section 3.9 Item II therefore needs to be revised into a concept- and dataset-specific statement, or the tables must be corrected.
  3. [§3.7, Table 15(a)] The specialist baselines 3D U-Net and EoFormer are reported with identical Dice values across all four MRI modalities (Flair, T1ce, T1, T2); for example, 3D U-Net shows 0.900/0.807/0.792 repeated verbatim in every column. Identical multi-modality results are implausible and suggest the scores were transcribed from a modality-independent source or copied incorrectly. Since Section 3.7 and Section 3.9 Item V use this comparison to claim that SAM 2 surpasses specialized models such as DRU-Net and 3D U-Net, the baseline values must be verified against the original publications and corrected; otherwise the superiority claim is unsupported.
minor comments (5)
  1. [§3.4.1] The OFS overlap threshold of 90% is introduced without any sensitivity analysis; a threshold sweep (for example, 50%, 75%, 90%) would clarify how much the everything-mode rankings depend on this arbitrary choice.
  2. [Tables 3–15] The meaning of "—" differs across tables: in Table 7 it indicates zero prediction accuracy per the PBD protocol, while in Table 9 it means the specialized model does not support that lesion type. A unified caption-level explanation would improve readability.
  3. [§3.6] The claim that SAM's image-based predictions "fluctuate frame by frame" is supported only by qualitative visualizations; a temporal consistency metric (for example, average inter-frame IoU of predicted masks) would make the comparison quantitative.
  4. [References [13, 14]] References [13] and [14] appear to be the same paper (Implicit Motion Handling for Video Camouflaged Object Detection) with different page ranges; this looks like a duplicate citation.
  5. [Prompt icons throughout] The prompt-type icons (for example, "/buromobelexperte", "/d⌢t-circle", "/border-s◎yle") render as odd symbols in the text; please use standard notation or clearly typeset glyphs in the final version.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the paper's claims are empirical measurements under clearly stated oracle-assisted evaluation protocols, not derivations that reduce to their own inputs.

full rationale

This is an empirical evaluation paper, not a derivation. The headline claims (box prompts generally most advantageous; SAM 2 worse on everything and point prompts; bidirectional inference helps 3D) are summaries of measured numbers under protocols that are explicitly described in Sec. 3.4. The everything-mode overlap filtering strategy (OFS) uses the ground-truth mask to retain predicted entities with >90% overlap, and the point-mode simulation uses GT-guided corrective clicks until IoU exceeds 0.9; the 3D anchor slice is chosen by largest GT foreground. These protocols condition on ground truth, which raises a validity question about whether the numbers describe zero-shot or realistic interactive behavior, but they do not make the results circular: the final predictions still depend on SAM/SAM2's raw outputs, and no parameter is fitted from the target quantity and then reported as a prediction. The paper also cites prior work by the same authors (Spider, [98]) for the context-dependent/context-independent taxonomy and for averaging protocols, but that citation is background framing rather than a load-bearing derivation of any experimental result. The identical baseline scores in Table 15a across modalities suggest a transcription concern, not circularity. Because no claimed result is equivalent by construction to its inputs and the evaluation methods are openly stated, the paper is self-contained as an empirical benchmark; concerns about oracle conditioning belong to experimental validity, not circularity.

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

This is an evaluation paper, so the ledger records protocol choices rather than fitted scientific parameters. The key bargain is that GT-derived prompts and GT-filtered outputs are the accepted way to measure SAMs, which conditions every headline number on oracle knowledge. No new particles, forces, or entities are introduced; the only new artifacts are evaluation protocols (propagation-based prompting, bidirectional inference, ICL mode, robustness perturbations). The CI/CD taxonomy is self-cited from the authors' own Spider paper.

free parameters (5)
  • OFS overlap threshold = 0.90 (90%)
    In automatic everything mode, only predicted entities whose overlap with the ground-truth mask exceeds 90% are merged into the final prediction (Sec. 3.4.1). All everything-mode scores in Tabs. 3 to 10 depend on this hand-chosen threshold.
  • Click stopping IoU and max clicks = IoU 0.9, max 6 clicks
    The point-prompt simulation iteratively adds corrective clicks until IoU exceeds 0.9 or six clicks are reached (Sec. 3.4.1). Point-mode results are conditional on this stopping rule.
  • Perturbation magnitudes = box 10% of shorter side, point 10 px, mask 5 erosion/dilation iterations
    Prompt robustness deltas in Tab. 16 are measured for these hand-chosen perturbation sizes; larger perturbations would change the reported sensitivity numbers.
  • ICL exemplar count = 20 training image-mask pairs
    SAM 2 in-context-learning results in Tab. 10 use the first 20 training samples per concept as prompts (Sec. 3.4.2); results scale with this choice.
  • 3D anchor slice selection = slice with largest foreground mask
    The bidirectional 3D protocol starts from the slice with the largest foreground region, which requires ground-truth knowledge (Sec. 3.4.4).
assumptions (5)
  • domain assumption The CI versus CD concept taxonomy of Barsalou, adopted via the authors' own Spider, is a valid organizing principle for segmentation benchmarks.
    Sec. 2.1 grounds the paper and selects the 11 evaluated concepts; the taxonomy originates in the authors' prior work [98].
  • domain assumption Ground-truth-derived prompts and GT-overlap filtering are an appropriate protocol for measuring segmentation capability.
    All interactive modes construct prompts from GT masks and the everything mode filters outputs by GT overlap (Sec. 3.4.1); the paper does not justify why this is preferable to a fully automatic protocol.
  • ad hoc to paper SAM 2's memory mechanism supports genuine in-context learning from 20 exemplar image-mask pairs.
    Sec. 3.4.2 passes 20 training image-mask pairs through the memory bank and interprets the result as ICL; no ablation varies exemplar count or checks whether gains come from mask leakage.
  • domain assumption Quoted specialist baselines from original papers are comparable to the SAM results measured here.
    Sec. 3.3 states baseline numbers are taken from original papers while SAMs use GT-derived prompts under this paper's protocol; differences could reflect protocol rather than capability.
  • ad hoc to paper The largest-foreground slice is the optimal starting anchor for 3D bidirectional propagation.
    Sec. 3.4.4 selects the anchor by largest foreground region without ablating alternative anchors.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Inspiring the Next Generation of Segment Anything Models: Comprehensively Evaluate SAM and SAM 2 with Diverse Prompts Towards Context-Dependent Concepts under Different Scenes." pith.science (2026). https://pith.science/paper/H72YTHPZ

@misc{pith2026241201240,
  author       = {Pith},
  title        = {Pith review of: Inspiring the Next Generation of Segment Anything Models: Comprehensively Evaluate SAM and SAM 2 with Diverse Prompts Towards Context-Dependent Concepts under Different Scenes},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/H72YTHPZ}},
  note         = {Machine review of arXiv:2412.01240}
}
read the original abstract

As large-scale foundation models trained on billions of image--mask pairs covering a vast diversity of scenes, objects, and contexts, SAM and its upgraded version, SAM~2, have significantly influenced multiple fields within computer vision. Leveraging such unprecedented data diversity, they exhibit strong open-world segmentation capabilities, with SAM~2 further enhancing these capabilities to support high-quality video segmentation. While SAMs (SAM and SAM~2) have demonstrated excellent performance in segmenting context-independent concepts like people, cars, and roads, they overlook more challenging context-dependent (CD) concepts, such as visual saliency, camouflage, industrial defects, and medical lesions. CD concepts rely heavily on global and local contextual information, making them susceptible to shifts in different contexts, which requires strong discriminative capabilities from the model. The lack of comprehensive evaluation of SAMs limits understanding of their performance boundaries, which may hinder the design of future models. In this paper, we conduct a thorough evaluation of SAMs on 11 CD concepts across 2D and 3D images and videos in various visual modalities within natural, medical, and industrial scenes. We develop a unified evaluation framework for SAM and SAM~2 that supports manual, automatic, and intermediate self-prompting, aided by our specific prompt generation and interaction strategies. We further explore the potential of SAM~2 for in-context learning and introduce prompt robustness testing to simulate real-world imperfect prompts. Finally, we analyze the benefits and limitations of SAMs in understanding CD concepts and discuss their future development in segmentation tasks.

Figures

Figures reproduced from arXiv: 2412.01240 by the authors.

Figure 1
Figure 1. Organization. (1) A unified evaluation framework for SAM and SAM 2; (2) Comprehensive evaluation for [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Visual comparison between context-independent concepts and context-dependent concepts. [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Architecture comparison for SAM and SAM 2. [PITH_FULL_IMAGE:figures/full_fig_p009_3.png] view at source ↗
Figures from the paper (12 more)
Figure 4
Figure 4. Figure 4: Visualization of salient object detection. From left to right: Image, Ground Truth, SAM ( [PITH_FULL_IMAGE:figures/full_fig_p014_4.png]
Figure 5
Figure 5. Figure 5: Visualization of camouflaged object detection. From left to right: Image, Ground Truth, SAM ( [PITH_FULL_IMAGE:figures/full_fig_p014_5.png]
Figure 6
Figure 6. Figure 6: Visualization of shadow detection. From left to right: Image, Ground Truth, SAM ( [PITH_FULL_IMAGE:figures/full_fig_p014_6.png]
Figure 7
Figure 7. Figure 7: Visualization of transparent object segmentation. From left to right: Image, Ground Truth, SAM ( [PITH_FULL_IMAGE:figures/full_fig_p014_7.png]
Figure 8
Figure 8. Figure 8: Visualization of power battery detection. From left to right: Image, Ground Truth, SAM ( [PITH_FULL_IMAGE:figures/full_fig_p015_8.png]
Figure 9
Figure 9. Figure 9: Visualization of surface anomaly detection. From left to right: Image, Ground Truth, SAM ( [PITH_FULL_IMAGE:figures/full_fig_p015_9.png]
Figure 10
Figure 10. Figure 10: Visualization of medical lesion segmentation (Lung, Breast, Skin, Polyp). From left to right: Image, Ground [PITH_FULL_IMAGE:figures/full_fig_p015_10.png]
Figure 11
Figure 11. Figure 11: Visualization of video polyp segmentation. [PITH_FULL_IMAGE:figures/full_fig_p016_11.png]
Figure 13
Figure 13. Figure 13: Visualization of video camouflaged object de [PITH_FULL_IMAGE:figures/full_fig_p017_13.png]
Figure 15
Figure 15. Figure 15: Visualization of 3D brain tumor segmentation. [PITH_FULL_IMAGE:figures/full_fig_p018_15.png]
Figure 17
Figure 17. Figure 17: Diverse prompt types. Potential directions include: I) Modalities like depth maps, infrared images, multispectral images, and X￾rays can provide valuable context beyond traditional RGB. These data types help models better understand scene and object structures, especi…
Figure 18
Figure 18. Figure 18: Four types of feature interaction between visual prompts and current target input. [PITH_FULL_IMAGE:figures/full_fig_p020_18.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

102 extracted references · 70 canonical work pages

  1. [1]

    Al-Dhabyani, M

    W. Al-Dhabyani, M. Gomaa, H. Khaled, and A. Fahmy. Dataset of breast ultrasound images. Data in Brief , 28:104863, 2020. 2, 8, 11

    2020

  2. [2]

    L. W. Barsalou. Context-independent and context- dependent information in concepts. Memory & Cog- nition, 10(1):82–93, 1982. 3, 5

    1982

  3. [3]

    Bergmann, M

    P. Bergmann, M. Fauser, D. Sattlegger, and C. Steger. Mvtec ad–a comprehensive real-world dataset for un- supervised anomaly detection. In Proceedings of IEEE Conference on Computer Vision and Pattern Recogni- tion, pages 9592–9600, 2019. 2, 8, 11, 13

    2019

  4. [4]

    Bernal, F

    J. Bernal, F. J. S´ anchez, G. Fern´ andez-Esparrach, D. Gil, C. Rodr ´ ıguez, and F. Vilari˜ no. Wm-dova maps for accurate polyp highlighting in colonoscopy: Valida- tion vs. saliency maps from physicians. Computerized Medical Imaging and Graphics , 43:99–111, 2015. 2, 8

    2015

  5. [5]

    Bideau and E

    P. Bideau and E. Learned-Miller. It’s moving! a proba- bilistic model for causal motion segmentation in moving camera videos. In Proceedings of European Conference on Computer Vision , pages 433–449, 2016. 2, 6, 12

    2016

  6. [6]

    V. I. Butoi, J. J. G. Ortiz, T. Ma, M. R. Sabuncu, J. Guttag, and A. V. Dalca. Universeg: Universal med- ical image segmentation. In Proceedings of IEEE Inter- national Conference on Computer Vision, pages 21438– 21451, 2023. 1, 5, 10, 11, 19

    2023

  7. [7]

    Caesar, J

    H. Caesar, J. Uijlings, and V. Ferrari. Coco-stuff: Thing and stuff classes in context. In Proceedings of IEEE Conference on Computer Vision and Pattern Recogni- tion, pages 1209–1218, 2018. 3

    2018

  8. [8]

    Carass, S

    A. Carass, S. Roy, A. Jog, J. L. Cuzzocreo, E. Magrath, A. Gherman, J. Button, J. Nguyen, F. Prados, C. H. Su- dre, et al. Longitudinal multiple sclerosis lesion segmen- tation: resource and challenge.NeuroImage, 148:77–102,

Show all 102 references
  1. [9]

    G. Chen, K. Han, and K.-Y. K. Wong. Tom-net: Learn- ing transparent object matting from a single image. In Proceedings of IEEE Conference on Computer Vision and Pattern Recognition, pages 9233–9241, 2018. 5

    2018

  2. [10]

    G. Chen, L. Li, Y. Dai, J. Zhang, and M. H. Yap. Aau- net: An adaptive attention u-net for breast lesions seg- mentation in ultrasound images. IEEE Transactions on Medical Imaging, 42(5):1289–1300, 2023. 11

    2023

  3. [11]

    T. Chen, L. Zhu, C. Ding, R. Cao, Y. Wang, Z. Li, L. Sun, P. Mao, and Y. Zang. Sam fails to segment anything?–sam-adapter: Adapting sam in underper- formed scenes: Camouflage, shadow, medical image seg- mentation, and more. arXiv preprint arXiv:2304.09148,

    arXiv

  4. [12]

    Z. Chen, Q. Xu, R. Cong, and Q. Huang. Global context-aware progressive aggregation network for salient object detection. In Proceedings of AAAI Con- ference on Artificial Intelligence , pages 10599–10606,

  5. [13]

    Cheng, H

    X. Cheng, H. Xiong, D. Fan, Y. Zhong, M. Harandi, T. Drummond, and Z. Ge. Implicit motion handling for video camouflaged object detection. In Proceedings of IEEE Conference on Computer Vision and Pattern Recognition, pages 13854–13863, 2022. 12

    2022

  6. [14]

    Cheng, H

    X. Cheng, H. Xiong, D.-P. Fan, Y. Zhong, M. Harandi, T. Drummond, and Z. Ge. Implicit motion handling for video camouflaged object detection. In Proceedings of IEEE Conference on Computer Vision and Pattern Recognition, pages 13864–13873, 2022. 2, 6, 12

    2022

  7. [15]

    C ¸ i¸ cek, A

    ¨O. C ¸ i¸ cek, A. Abdulkadir, S. S. Lienkamp, T. Brox, and O. Ronneberger. 3d u-net: Learning dense volumetric segmentation from sparse annotation. In S. Ourselin, L. Joskowicz, M. R. Sabuncu, G. B. ¨Unal, and W. M. W. III, editors, Proceedings of International Conference on ...

    2016

  8. [16]

    Codella, V

    N. Codella, V. Rotemberg, P. Tschandl, M. E. Celebi, S. Dusza, D. Gutman, B. Helba, A. Kalloo, K. Liopy- ris, M. Marchetti, et al. Skin lesion analysis toward melanoma detection 2018: A challenge hosted by the international skin imaging collaboration (isic). arXiv preprint arX...

    2018 arXiv

  9. [17]

    R. Cong, H. Yang, Q. Jiang, W. Gao, H. Li, C. Wang, Y. Zhao, and S. Kwong. Bcs-net: Boundary, context, and semantic for automatic covid-19 lung infection seg- mentation from ct images. IEEE Transactions on In- strumentation and Measurement , 71:1–11, 2022. 5

    2022

  10. [18]

    Cordts, M

    M. Cordts, M. Omran, S. Ramos, T. Rehfeld, M. En- zweiler, R. Benenson, U. Franke, S. Roth, and B. Schiele. The cityscapes dataset for semantic urban scene under- standing. In Proceedings of IEEE Conference on Com- puter Vision and Pattern Recognition, pages 3213–3223,

  11. [19]

    Deng and X

    H. Deng and X. Li. Anomaly detection via reverse distillation from one-class embedding. In Proceedings of IEEE Conference on Computer Vision and Pattern Recognition, pages 9727–9736, 2022. 11

    2022

  12. [20]

    Fan, M.-M

    D.-P. Fan, M.-M. Cheng, Y. Liu, T. Li, and A. Borji. Structure-measure: A new way to evaluate foreground maps. In Proceedings of IEEE International Conference on Computer Vision , pages 4548–4557, 2017. 8

    2017

  13. [21]

    Fan, G.-P

    D.-P. Fan, G.-P. Ji, M.-M. Cheng, and L. Shao. Con- cealed object detection. IEEE Transactions on Pattern Analysis and Machine Intelligence , 44(10):6024–6042,

  14. [22]

    Fan, G.-P

    D.-P. Fan, G.-P. Ji, G. Sun, M.-M. Cheng, J. Shen, and L. Shao. Camouflaged object detection. In Proceedings of IEEE Conference on Computer Vision and Pattern Recognition, pages 2777–2787, 2020. 2, 6, 11

    2020

  15. [23]

    Fan, G.-P

    D.-P. Fan, G.-P. Ji, P. Xu, M.-M. Cheng, C. Sakaridis, and L. Van Gool. Advances in deep concealed scene understanding. Visual Intelligence , 1(1):16, 2023. 2

    2023

  16. [24]

    Fan, G.-P

    D.-P. Fan, G.-P. Ji, T. Zhou, G. Chen, H. Fu, J. Shen, and L. Shao. Pranet: Parallel reverse attention net- work for polyp segmentation. In Proceedings of Inter- national Conference on Medical Image Computing and Computer-Assisted Intervention , pages 263–273, 2020. 5, 11

    2020

  17. [25]

    D.-P. Fan, W. Wang, M.-M. Cheng, and J. Shen. Shift- ing more attention to video salient object detection. In Proceedings of IEEE Conference on Computer Vision and Pattern Recognition, pages 8554–8564, 2019. 2, 6, 12

    2019

  18. [26]

    D.-P. Fan, T. Zhou, G.-P. Ji, Y. Zhou, G. Chen, H. Fu, J. Shen, and L. Shao. Inf-net: Automatic covid-19 lung infection segmentation from ct images. IEEE Transac- tions on Medical Imaging , 39(8):2626–2637, 2020. 2, 8, 11

    2020

  19. [27]

    K. Fan, C. Wang, Y. Wang, C. Wang, R. Yi, and L. Ma. Rfenet: Towards reciprocal feature evolution for glass segmentation. In Proceedings of International Joint Conference on Artificial Intelligence , pages 717–725,

  20. [28]

    Hashemi, M

    M. Hashemi, M. Akhbari, and C. Jutten. Delve into mul- tiple sclerosis (MS) lesion exploration: A modified at- tention u-net for MS lesion segmentation in brain MRI. 22 Xiaoqi Zhao 1 et al. Computers in Biology and Medicine , 145:105402, 2022. 12

    2022

  21. [29]

    H. He, X. Li, G. Cheng, J. Shi, Y. Tong, G. Meng, V. Prinet, and L. Weng. Enhanced boundary learn- ing for glass-like object segmentation. In Proceedings of IEEE International Conference on Computer Vision , pages 15839–15848, 2021. 11

    2021

  22. [30]

    J. Hu, Y. Yang, X. Guo, B. Peng, H. Huang, and T. Ma. Decor-net: A covid-19 lung infection segmentation net- work improved by emphasizing low-level features and decorrelating features. In IEEE International Sympo- sium on Biomedical Imaging , pages 1–5, 2023. 11

    2023

  23. [31]

    D. Jha, P. H. Smedsrud, M. A. Riegler, P. Halvorsen, T. d. Lange, D. Johansen, and H. D. Johansen. Kvasir- seg: A segmented polyp dataset. In Proceedings of In- ternational Conference on Multimedia Modeling , pages 451–462, 2020. 2, 8

    2020

  24. [32]

    Ji, Y.-C

    G.-P. Ji, Y.-C. Chou, D.-P. Fan, G. Chen, H. Fu, D. Jha, and L. Shao. Progressively normalized self-attention network for video polyp segmentation. In Proceedings of International Conference on Medical Image Com- puting and Computer-Assisted Intervention, pages 142– 152, 2021....

    2021

  25. [33]

    Ji, D.-P

    G.-P. Ji, D.-P. Fan, P. Xu, M.-M. Cheng, B. Zhou, and L. Van Gool. Sam struggles in concealed scenes– empirical study on” segment anything”. arXiv preprint arXiv:2304.06022, 2023. 2, 3, 6

    2023 arXiv

  26. [34]

    W. Ji, J. Li, Q. Bi, T. Liu, W. Li, and L. Cheng. Seg- ment anything is not always perfect: An investigation of sam on different real-world applications. Machine Intelligence Research, 21:617–630, 2024. 2, 6

    2024

  27. [35]

    L. Ke, M. Ye, M. Danelljan, Y. Liu, Y.-W. Tai, C.-K. Tang, and F. Yu. Segment anything in high quality. In Proceedings of International Conference and Workshop on Neural Information Processing Systems , 2024. 1, 5

    2024

  28. [36]

    D. P. Kingma. Adam: A method for stochastic opti- mization. In Proceedings of International Conference on Learning Representations, 2015. 20

    2015

  29. [37]

    Kirillov, E

    A. Kirillov, E. Mintun, N. Ravi, H. Mao, C. Rolland, L. Gustafson, T. Xiao, S. Whitehead, A. C. Berg, W.- Y. Lo, et al. Segment anything. In Proceedings of IEEE International Conference on Computer Vision , pages 4015–4026, 2023. 1, 19

    2023

  30. [38]

    Lachmann and C

    T. Lachmann and C. Van Leeuwen. Individual pattern representations are context independent, but their col- lectiverepresentation is context dependent. The Quar- terly Journal of Experimental Psychology Section A , 58:1265–1294, 2005. 5

    2005

  31. [39]

    Lamdouar, C

    H. Lamdouar, C. Yang, W. Xie, and A. Zisserman. Be- trayed by motion: Camouflaged object discovery via mo- tion segmentation. 2020. 2

    2020

  32. [40]

    T.-N. Le, T. V. Nguyen, Z. Nie, M.-T. Tran, and A. Sug- imoto. Anabranch network for camouflaged object seg- mentation. Computer Vision and Image Understand- ing, 184:45–56, 2019. 2, 6, 11

    2019

  33. [41]

    Li and Y

    G. Li and Y. Yu. Visual saliency based on multiscale deep features. In Proceedings of IEEE Conference on Computer Vision and Pattern Recognition, pages 5455– 5463, 2015. 2, 6, 11

    2015

  34. [42]

    Y. Li, X. Hou, C. Koch, J. M. Rehg, and A. L. Yuille. The secrets of salient object segmentation. In Proceed- ings of IEEE Conference on Computer Vision and Pat- tern Recognition, pages 280–287, 2014. 2, 6, 11

    2014

  35. [43]

    Lian and H

    S. Lian and H. Li. Evaluation of segment anything model 2: The role of sam2 in the underwater environ- ment. arXiv preprint arXiv:2408.02924 , 2024. 2, 3, 6

    2024 arXiv

  36. [44]

    S. Lian, Z. Zhang, H. Li, W. Li, L. T. Yang, S. Kwong, and R. Cong. Diving into underwater: Segment anything model guided underwater salient instance segmentation and a large-scale dataset. In Proceedings of Interna- tional Conference on Machine Learning , 2024. 2

    2024

  37. [45]

    J. Lin, Z. He, and R. W. Lau. Rich context aggregation with reflection prior for glass surface detection. In Pro- ceedings of IEEE Conference on Computer Vision and Pattern Recognition, pages 13415–13424, 2021. 5

    2021

  38. [46]

    N. Liu, K. Nan, W. Zhao, X. Yao, and J. Han. Learning complementary spatial–temporal transformer for video salient object detection. IEEE Transactions on Neu- ral Networks and Learning Systems, 35(8):10663–10673,

  39. [47]

    W. Liu, X. Shen, C.-M. Pun, and X. Cun. Explicit vi- sual prompting for low-level structure segmentations. In Proceedings of IEEE Conference on Computer Vision and Pattern Recognition, pages 19434–19445, 2023. 5

    2023

  40. [48]

    Y. Liu, M. Zhu, H. Li, H. Chen, X. Wang, and C. Shen. Matcher: Segment anything with one shot using all-purpose feature matching. arXiv preprint arXiv:2305.13310, 2023. 1

    2023 arXiv

  41. [49]

    X. Lu, Y. Cao, S. Liu, C. Long, Z. Chen, X. Zhou, Y. Yang, and C. Xiao. Video shadow detection via spatio-temporal interpolation consistency training. In Proceedings of IEEE Conference on Computer Vision and Pattern Recognition, pages 3116–3125, 2022. 2, 7, 11, 12

    2022

  42. [50]

    X. Lu, Y. Cao, S. Liu, C. Long, Z. Chen, X. Zhou, Y. Yang, and C. Xiao. Video shadow detection via spatio-temporal interpolation consistency training. In Proceedings of IEEE Conference on Computer Vision and Pattern Recognition, pages 3106–3115, 2022. 12

    2022

  43. [51]

    Z. Luo, N. Liu, W. Zhao, X. Yang, D. Zhang, D.-P. Fan, F. Khan, and J. Han. Vscode: General visual salient and camouflaged object detection with 2d prompt learning. In Proceedings of IEEE Conference on Computer Vi- sion and Pattern Recognition, pages 17169–17180, 2024. 5

    2024

  44. [52]

    Y. Lv, J. Zhang, Y. Dai, A. Li, B. Liu, N. Barnes, and D.-P. Fan. Simultaneously localize, segment and rank the camouflaged objects. In Proceedings of IEEE Con- ference on Computer Vision and Pattern Recognition , pages 11591–11601, 2021. 2, 6, 11

    2021

  45. [53]

    J. Ma, Y. He, F. Li, L. Han, C. You, and B. Wang. Segment anything in medical images. Nature Commu- nications, 15(1):654, 2024. 1

    2024

  46. [54]

    Margolin, L

    R. Margolin, L. Zelnik-Manor, and A. Tal. How to eval- uate foreground maps? In Proceedings of IEEE Con- ference on Computer Vision and Pattern Recognition , pages 248–255, 2014. 8

    2014

  47. [55]

    Martial, D

    C. Martial, D. Stawarczyk, and A. D’Argembeau. Neural correlates of context-independent and context- dependent self-knowledge. Brain and Cognition , 125:23–31, 2018. 5

    2018

  48. [56]

    Mehta, A

    R. Mehta, A. Filos, U. Baid, C. Sako, R. McKinley, M. Rebsamen, K. Datwyler, R. Meier, P. Radojewski, G. K. Murugesan, et al. Qu-brats: Miccai brats 2020 challenge on quantifying uncertainty in brain tumor segmentation-analysis of ranking scores and benchmark- ing results. arX...

    2020 arXiv

  49. [57]

    Mishra, R

    P. Mishra, R. Verk, D. Fornasier, C. Piciarelli, and G. L. Foresti. Vt-adl: A vision transformer network for im- age anomaly detection and localization. In Proceedings of International Symposium on Industrial Electronics , pages 01–06, 2021. 2, 8, 11 Comprehensive Evaluation o...

    2021

  50. [58]

    Y. Pang, X. Zhao, T.-Z. Xiang, L. Zhang, and H. Lu. Zoomnext: A unified collaborative pyramid network for camouflaged object detection. IEEE Transactions on Pattern Analysis and Machine Intelligence , 2024. 5, 6, 11, 12

    2024

  51. [59]

    Perazzi, J

    F. Perazzi, J. Pont-Tuset, B. McWilliams, L. Van Gool, M. Gross, and A. Sorkine-Hornung. A benchmark dataset and evaluation methodology for video object segmentation. In Proceedings of IEEE Conference on Computer Vision and Pattern Recognition , pages 724– 732, 2016. 2, 6, 12, 13

    2016

  52. [60]

    X. Qin, H. Dai, X. Hu, D.-P. Fan, L. Shao, and L. Van Gool. Highly accurate dichotomous image seg- mentation. In ECCV, pages 38–56, 2022. 2

    2022

  53. [61]

    N. Ravi, V. Gabeur, Y.-T. Hu, R. Hu, C. Ryali, T. Ma, H. Khedr, R. R¨ adle, C. Rolland, L. Gustafson, et al. Sam 2: Segment anything in images and videos. arXiv preprint arXiv:2408.00714, 2024. 1, 19

    2024 arXiv

  54. [62]

    K. Roth, L. Pemula, J. Zepeda, B. Sch¨ olkopf, T. Brox, and P. V. Gehler. Towards total recall in industrial anomaly detection. In Proceedings of IEEE Conference on Computer Vision and Pattern Recognition , pages 14298–14308, 2022. 11

    2022

  55. [63]

    J. Ruan, S. Xiang, M. Xie, T. Liu, and Y. Fu. Malunet: A multi-attention and light-weight unet for skin le- sion segmentation. In IEEE International Conference on Bioinformatics and Biomedicine , pages 1150–1156,

  56. [64]

    J. Ruan, M. Xie, J. Gao, T. Liu, and Y. Fu. Ege-unet: An efficient group enhanced unet for skin lesion segmen- tation. In Proceedings of International Conference on Medical Image Computing and Computer-Assisted In- tervention, pages 481–490, 2023. 11

    2023

  57. [65]

    A. N. Rumaksari, S. Sumpeno, and A. D. Wibawa. Background subtraction using spatial mixture of gaus- sian model with dynamic shadow filtering. In Proceed- ings of International Seminar on Intelligent Technology and Its Applications , pages 296–301, 2017. 5

    2017

  58. [66]

    Sarica, D

    B. Sarica, D. Z. Seker, and B. Bayram. A dense residual u-net for multiple sclerosis lesions segmentation from multi-sequence 3d MR images. International Journal of Medical Informatics , 170:104965, 2023. 12

    2023

  59. [67]

    D. She, Y. Zhang, Z. Zhang, H. Li, Z. Yan, and X. Sun. Eoformer: Edge-oriented transformer for brain tumor segmentation. In Proceedings of International Con- ference on Medical Image Computing and Computer- Assisted Intervention, pages 333–343, 2023. 12

    2023

  60. [68]

    Silva, A

    J. Silva, A. Histace, O. Romain, X. Dray, and B. Granado. Toward embedded detection of polyps in wce images for early diagnosis of colorectal cancer. International Journal of Computer Assisted Radiology and Surgery, 9:283–293, 2014. 2, 8

    2014

  61. [69]

    Skurowski, H

    P. Skurowski, H. Abdulameer, J. B laszczyk, T. Depta, A. Kornacki, and P. Kozie l. Animal camouflage analysis: Chameleon database. Unpublished Manuscript, 2018. 2

    2018

  62. [70]

    J. Sun, K. Xu, Y. Pang, L. Zhang, H. Lu, G. P. Hancke, and R. W. H. Lau. Adaptive illumination mapping for shadow detection in raw images. InProceedings of IEEE International Conference on Computer Vision , pages 12663–12672, 2023. 11

    2023

  63. [71]

    Tajbakhsh, S

    N. Tajbakhsh, S. R. Gurudu, and J. Liang. Automated polyp detection in colonoscopy videos using shape and context information. IEEE Transactions on Medical Imaging, 35(2):630–644, 2015. 2, 8

    2015

  64. [72]

    F. Tang, L. Wang, C. Ning, M. Xian, and J. Ding. Cmu- net: A strong convmixer-based medical ultrasound im- age segmentation network. In IEEE International Sym- posium on Biomedical Imaging , pages 1–5, 2023. 11

    2023

  65. [73]

    Tang and B

    L. Tang and B. Li. Evaluating sam2’s role in cam- ouflaged object detection: From sam to sam2. arXiv preprint arXiv:2407.21596, 2024. 2

    2024 arXiv

  66. [74]

    Z. Tu, T. Xia, C. Li, X. Wang, Y. Ma, and J. Tang. Rgb-t image saliency detection via collaborative graph learning. IEEE Transactions on Multimedia, 22(1):160– 173, 2019. 2

    2019

  67. [75]

    V´ azquez, J

    D. V´ azquez, J. Bernal, F. J. S´ anchez, G. Fern´ andez- Esparrach, A. M. L´ opez, A. Romero, M. Drozdzal, and A. Courville. A benchmark for endoluminal scene seg- mentation of colonoscopy images. Journal of Healthcare Engineering, 2017(1):4037190, 2017. 2, 8

    2017

  68. [76]

    T. F. Y. Vicente, M. Hoai, and D. Samaras. Leave- one-out kernel optimization for shadow detection. In Proceedings of IEEE International Conference on Com- puter Vision , pages 3388–3396, 2015. 8

    2015

  69. [77]

    T. F. Y. Vicente, L. Hou, C.-P. Yu, M. Hoai, and D. Samaras. Large-scale training of shadow detectors with noisily-annotated shadow examples. In Proceedings of European Conference on Computer Vision , pages 816–832, 2016. 2, 7, 11

    2016

  70. [78]

    J. Wang, X. Li, and J. Yang. Stacked conditional gen- erative adversarial networks for jointly learning shadow detection and shadow removal. In Proceedings of IEEE Conference on Computer Vision and Pattern Recogni- tion, pages 1788–1797, 2018. 2, 7, 11

    2018

  71. [79]

    L. Wang, H. Lu, Y. Wang, M. Feng, D. Wang, B. Yin, and X. Ruan. Learning to detect salient objects with image-level supervision. In Proceedings of IEEE Con- ference on Computer Vision and Pattern Recognition , pages 136–145, 2017. 2, 6, 11, 13

    2017

  72. [80]

    M. Wang, Y. Feng, Y. Tang, T. Zhang, Y. Liang, and C. Lv. Global-local medical sam adaptor based on full adaption. arXiv preprint arXiv:2409.17486 , 2024. 1

    2024

  73. [81]

    X. Wang, W. Wang, Y. Cao, C. Shen, and T. Huang. Im- ages speak in images: A generalist painter for in-context visual learning. In Proceedings of IEEE Conference on Computer Vision and Pattern Recognition, pages 6830– 6839, 2023. 1

    2023

  74. [82]

    X. Wang, X. Zhang, Y. Cao, W. Wang, C. Shen, and T. Huang. Seggpt: Towards segmenting everything in context. In Proceedings of IEEE International Confer- ence on Computer Vision , pages 1130–1140, 2023. 1, 5, 10, 11, 19

    2023

  75. [83]

    Y. Wang, R. Wang, X. Fan, T. Wang, and X. He. Pixels, regions, and objects: Multiple enhancement for salient object detection. In Proceedings of IEEE Conference on Computer Vision and Pattern Recognition , pages 10031–10040, 2023. 11

    2023

  76. [84]

    J. Wei, Y. Hu, S. Cui, S. K. Zhou, and Z. Li. Weakpolyp: You only look bounding box for polyp segmentation. In Proceedings of International Conference on Medical Image Computing and Computer-Assisted Intervention, pages 757–766, 2023. 11

    2023

  77. [85]

    J. Wu, W. Ji, Y. Liu, H. Fu, M. Xu, Y. Xu, and Y. Jin. Medical sam adapter: Adapting segment any- thing model for medical image segmentation. arXiv preprint arXiv:2304.12620, 2023. 1

    2023 arXiv

  78. [86]

    Y. Wu, Y. Liu, L. Zhang, M. Cheng, and B. Ren. EDN: salient object detection via extremely-downsampled net- work. IEEE Transactions on Image Processing , pages 3125–3136, 2022. 11

    2022

  79. [87]

    E. Xie, W. Wang, W. Wang, M. Ding, C. Shen, and P. Luo. Segmenting transparent objects in the wild. 24 Xiaoqi Zhao 1 et al. In Proceedings of European Conference on Computer Vision, pages 696–711, 2020. 2, 7, 11

    2020

  80. [88]

    H. Xing, S. Gao, Y. Wang, X. Wei, H. Tang, and W. Zhang. Go closer to see better: Camouflaged ob- ject detection via object area amplification and figure- ground conversion. IEEE Transactions on Circuits and Systems for Video Technology, 33(10):5444–5457, 2023. 6, 11

    2023

  81. [89]

    Q. Yan, L. Xu, J. Shi, and J. Jia. Hierarchical saliency detection. In Proceedings of IEEE Conference on Com- puter Vision and Pattern Recognition, pages 1155–1162,

  82. [90]

    C. Yang, L. Zhang, H. Lu, X. Ruan, and M.-H. Yang. Saliency detection via graph-based manifold ranking. In Proceedings of IEEE Conference on Computer Vision and Pattern Recognition, pages 3166–3173, 2013. 2, 6, 11

    2013

  83. [91]

    H. Yang, T. Wang, X. Hu, and C. Fu. SILT: shadow- aware iterative label tuning for learning to detect shad- ows from noisy labels. In Proceedings of IEEE Interna- tional Conference on Computer Vision , pages 12641– 12652, 2023. 11

    2023

  84. [92]

    X. Yuan, G. Cheng, K. Yan, Q. Zeng, and J. Han. Small object detection via coarse-to-fine proposal generation and imitation learning. In Proceedings of IEEE Inter- national Conference on Computer Vision , pages 6317– 6327, 2023. 11

    2023

  85. [93]

    Zhang, D.-P

    J. Zhang, D.-P. Fan, Y. Dai, X. Yu, Y. Zhong, N. Barnes, and L. Shao. Rgb-d saliency detection via cascaded mu- tual information minimization. In Proceedings of IEEE International Conference on Computer Vision , pages 4338–4347, 2021. 2

    2021

  86. [94]

    Zhang, P

    R. Zhang, P. Lai, X. Wan, D. Fan, F. Gao, X. Wu, and G. Li. Lesion-aware dynamic kernel for polyp segmen- tation. In Proceedings of International Conference on Medical Image Computing and Computer-Assisted In- tervention, pages 99–109, 2022. 11

    2022

  87. [95]

    X. Zhao, H. Jia, Y. Pang, L. Lv, F. Tian, L. Zhang, W. Sun, and H. Lu. M 2snet: Multi-scale in multi-scale subtraction network for medical image segmentation. arXiv preprint arXiv:2303.10894 , 2023. 12

    2023 arXiv

  88. [96]

    X. Zhao, H. Liang, P. Li, G. Sun, D. Zhao, R. Liang, and X. He. Motion-aware memory network for fast video salient object detection. IEEE Transactions on Image Processing, 33:709–721, 2024. 12

    2024

  89. [97]

    X. Zhao, Y. Pang, Z. Chen, Q. Yu, L. Zhang, H. Liu, J. Zuo, and H. Lu. Towards automatic power battery de- tection: New challenge benchmark dataset and baseline. In Proceedings of IEEE Conference on Computer Vi- sion and Pattern Recognition, pages 22020–22029, 2024. 2, 8, 11

    2024

  90. [98]

    X. Zhao, Y. Pang, W. Ji, B. Sheng, J. Zuo, L. Zhang, and H. Lu. Spider: A unified framework for context- dependent concept understanding. In Proceedings of International Conference on Machine Learning , pages 60906–60926, 2024. 1, 3, 5, 8, 10, 11, 19

    2024

  91. [99]

    X. Zhao, Y. Pang, L. Zhang, H. Lu, and L. Zhang. To- wards diverse binary segmentation via a simple yet gen- eral gated network. International Journal of Computer Vision, 132:4157–4234, 2024. 5

    2024

  92. [100]

    X. Zhao, L. Zhang, and H. Lu. Automatic polyp seg- mentation via multi-scale subtraction network. In Pro- ceedings of International Conference on Medical Image Computing and Computer-Assisted Intervention , pages 120–130, 2021. 5

    2021

  93. [101]

    T. Zhou, Y. Zhang, Y. Zhou, Y. Wu, and C. Gong. Can sam segment polyps? arXiv preprint arXiv:2304.07583,

    arXiv

  94. [102]

    Y. Zou, J. Jeong, L. Pemula, D. Zhang, and O. Dabeer. Spot-the-difference self-supervised pre- training for anomaly detection and segmentation. In Proceedings of European Conference on Computer Vi- sion, pages 392–408, 2022. 2, 8

    2022

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.