Pith. sign in

REVIEW 3 major objections 4 minor 61 references

MSRNet is a multi-scale recursive network that detects small and multiple camouflaged objects; the paper reports state-of-the-art results on two COD benchmarks and second-best on the other two.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

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

T0 review · deepseek-v4-flash

2026-08-03 21:55 UTC pith:ISIWNEFZ

load-bearing objection Very likely a re-skin of ZoomNeXt — the parameter counts match to 0.001M across every setting, so the novelty claim needs verification. the 3 major comments →

arxiv 2511.12810 v2 pith:ISIWNEFZ submitted 2025-11-16 cs.CV cs.AIeess.IV

MSRNet: A Multi-Scale Recursive Network for Camouflaged Object Detection

Leena Alghamdi , Muhammad Usman , Hafeez Anwar , Abdul Bais , Saeed Anwar This is my paper
classification cs.CV cs.AIeess.IV
keywords camouflaged object detectionmulti-scale feature learningrecursive-feedback decodingPyramid Vision TransformerAttention-Based Scale Integration UnitMulti-Granularity Fusion Unitsmall object detectionCOD benchmarks
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

MSRNet is a transformer-based architecture for camouflaged object detection, the task of finding objects that blend into the background. The paper argues that combining multi-scale inputs (1x, 1.5x, and 2x), attention-based scale integration, and a recursive-feedback decoder lets the model detect small, tiny, and multiple camouflaged objects more reliably than 20 prior methods. It reports top results on two of the four benchmark test sets and second-best results on the other two. This matters because COD is relevant to medical imaging, search-and-rescue, agriculture, and defect detection, and small or multiple objects are exactly the cases where earlier models most often fail. Ablations support the claim by showing that larger input scales and the recursive-feedback strategy both help, while dense aggregation of raw multi-resolution features hurts.

Core claim

The paper's central claim is that camouflaged object detection can be improved by processing the input at three scales and passing the features through a recursive-feedback decoder. MSRNet feeds the original image and 1.5x and 2x enlarged versions through a Pyramid Vision Transformer encoder; an Attention-Based Scale Integration Unit (ABSIU) merges same-resolution features across scales; and Multi-Granularity Fusion Units (MGFUs) in the decoder refine each resolution while receiving feedback from all lower-resolution stages. On this design, the paper reports the best published results on two of the four standard COD benchmarks, second-best on the other two, and qualitative results showing re

What carries the argument

The recursive-feedback decoding strategy is the load-bearing mechanism: it combines progressive decoding with an advanced feedback loop that takes lower-resolution feature maps and feeds them into every subsequent higher-resolution refinement stage, preserving global context. This is implemented with Multi-Granularity Fusion Units (MGFUs), which split features into channel groups, propagate information between groups, and gate important channels. On the encoding side, the Attention-Based Scale Integration Unit (ABSIU) merges features from the three input scales for each resolution using multi-head spatial attention. The paper attributes detection of small objects mainly to the larger input s

Load-bearing premise

The load-bearing premise is that the final configuration is genuinely better and not merely the one that scored highest on the particular test benchmarks used for model selection; if that configuration does not carry over to a held-out split, the reported margins over prior methods may be optimistic.

What would settle it

Run MSRNet with a fixed configuration chosen only on a held-out validation subset, then evaluate on the official test splits; if the margins over published prior methods shrink or the ranking changes, the paper's SOTA claim is an artifact of test-set model selection.

Watch this falsifier — get emailed when new claim-graph text bears on it.

If this is right

  • If the reported results hold, a multi-scale input pyramid plus recursive feedback is a viable recipe for COD, beating prior specialized methods on standard benchmarks without extra training data.
  • The ablation's largest single gain comes from raising the input scales to 1.0x, 1.5x, and 2.0x, suggesting that resolution itself is a major driving factor in small-object detection.
  • The recursive-feedback decoder adds a small but consistent gain over progressive-only decoding, while dense aggregation of unrefined multi-resolution features hurts, indicating that the way features are combined matters as much as which features are combined.
  • Because the model has roughly the same parameter count as comparable same-backbone competitors, the paper attributes the performance difference to the multi-scale and recursive design rather than to a larger network.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • Editorial inference: An independent test that holds the decoder fixed and varies only the input scales (1x-only versus 1x-plus-enlargements) would quantify how much of the SOTA margin is actually due to multi-scale input versus the recursive architecture.
  • Editorial inference: The recursive-feedback pattern, where lower-resolution context is injected into every higher-resolution refinement stage, is a general mechanism that could transfer to other dense-prediction tasks with strong contextual demands, such as salient-object detection, polyp segmentation, or defect localization.
  • Editorial inference: The paper's ablations are all reported on the official test benchmarks, so the winning configuration (PVTv2-B4, 384x384, scales 1.0/1.5/2.0) is the one that scored highest on those exact test sets; a validation-set-based model selection would make the headline comparison more decisive.
  • Editorial inference: A direct way to test the ABSIU's contribution is to ablate it with plain averaging of the scale features; if the attention mechanism adds little, the gains attributed to scale integration may instead come mostly from the enlarged input scales.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 4 minor

Summary. The paper proposes MSRNet, a multi-scale recursive network for camouflaged object detection (COD). It extracts features from three input scales (1.0x, 1.5x, 2.0x) with a PVTv2 backbone, merges them with Attention-Based Scale Integration Units (ABSIUs), and refines features in a decoder with Multi-Granularity Fusion Units (MGFUs) under a recursive-feedback decoding strategy. The authors report state-of-the-art or near-state-of-the-art results on CAMO, CHAMELEON, COD10K, and NC4K, and claim improvements especially for small and multiple camouflaged objects. The paper is clearly written and includes extensive comparisons with 20 prior methods, ablation studies, qualitative examples, and a public code/weight release.

Significance. If the proposed architecture is genuinely distinct and the empirical claims are statistically sound, the work would be a useful contribution to COD: the combination of multi-scale input, attention-based scale integration, and recursive-feedback refinement is a reasonable design direction, and the reported gains on four standard benchmarks are meaningful. The authors also provide comprehensive ablations and make code/weights available. However, the significance hinges on two unresolved issues: (1) the reported parameter counts are essentially identical to ZoomNeXt for every configuration, casting doubt on whether ABSIU/MGFU are actually different from ZoomNeXt's modules; and (2) the reported final model appears to be selected on the test sets, which can inflate results. The paper's concrete strengths—benchmark breadth, modular ablations, and public release—do not compensate for these threats unless properly addressed.

major comments (3)
  1. [Table 2, Section 3.2–3.3] The parameter counts of MSRNet match ZoomNeXt to within 0.001M for every backbone/input configuration (e.g., PVTv2-B2 352×352: 28.180M vs 28.181M; PVTv2-B4 384×384: 65.373M vs 65.374M; EfficientNet-B4: 21.381M vs 21.381M). This is statistically implausible for two independently designed architectures with different non-trivial modules, especially since ABSIU and MGFU are described as different from ZoomNeXt's scale-integration and feature-refinement units. The authors must supply a module-by-module parameter breakdown, official code/checkpoints, or a proof of parameter-equivalence; absent that, the reported results may simply reproduce ZoomNeXt under a different name and the central novelty claim collapses.
  2. [Section 4.3, Table 3] The ablation study selects the final model (M9) based on performance on the CAMO, CHAMELEON, COD10K, and NC4K test sets. The paper evaluates multiple decoders, input sizes, backbones, and scale sets directly on these test benchmarks, then reports the best-performing combination as the final model in Table 2. This is model selection on the test data, which biases the reported SOTA numbers optimistically. A held-out validation split or a nested validation protocol is needed, or the authors should explicitly acknowledge the selection procedure and quantify its optimism (e.g., via cross-validation). This issue is load-bearing for the headline claim of state-of-the-art results.
  3. [Section 4.2, Table 2] All competitor results are imported from published papers, and no error bars or significance tests are provided. Many differences between MSRNet and the closest competitors are at the third decimal place (e.g., CAMO S_m 0.888 vs 0.888; CAMO F_β^ω 0.861 vs 0.859; CHAMELEON S_m 0.932 vs 0.929). Without error bars, significance tests, or at least re-run baselines under identical training conditions, the claimed 'state-of-the-art' status is not established. The authors should report variance over at least a few seeds or provide statistical tests for the key comparisons.
minor comments (4)
  1. [Abstract and Section 4.2] The text says 'state-of-the-art results on the CAMO10K and NC4K datasets' and then 'ranks second on both CAMO and CHAMELEON.' If 'CAMO10K' refers to CAMO, this is contradictory; if it refers to COD10K, the name is inconsistent. Please clarify.
  2. [Table 3] The baseline B0 is called 'fully equipped' in the text but then M1 adds RFD, which is part of the proposed decoder. Please make clear which components are in B0 and how the Δ values are computed (averaged over which metrics?).
  3. [Section 4.2] The text switches between 'CNN-based', 'vision-based', and 'vision transformer based' when describing model categories; please be consistent, especially because 'vision-based' is vague.
  4. [Figure 3 caption] Typo: 'The the overall architecture' should be 'The overall architecture.'

Circularity Check

0 steps flagged

No significant circularity: the paper's claims are empirical benchmark comparisons with no derivation chain that reduces to its inputs.

full rationale

The paper makes an empirical performance claim: MSRNet is evaluated on four standard COD benchmarks against published SOTA methods, and its reported numbers are measurements, not derivations. There is no equation or fitted parameter from which the benchmark scores are mathematically forced. The UAL loss is explicitly credited to ZoomNeXt [52] and is an ordinary training loss, not a self-citation and not load-bearing for the claimed novelty. The ablation study (Table 3) selects configurations on the test benchmarks, which is a statistical-optimism/correctness concern, not a circularity step, because the final numbers are not predictions from a pre-specified model but selected results. The striking parameter-count identity with ZoomNeXt in Table 2 is likewise an external novelty/integrity concern (possible renamed architecture), not a circular derivation: no quote or equation in the paper exhibits MSRNet reducing to ZoomNeXt by construction. Under the hard rule requiring a quoted specific reduction, no circular step is established.

Axiom & Free-Parameter Ledger

5 free parameters · 4 axioms · 0 invented entities

MSRNet introduces no new physical entities or external theoretical posits. Its assumptions are standard machine-learning transfer and benchmarking assumptions, and its free parameters are architectural and training hyperparameters selected largely through ablation on the evaluation benchmarks.

free parameters (5)
  • Input scales {1.0x, 1.5x, 2.0x} = 1.0x, 1.5x, 2.0x
    Selected via ablation in Section 4.3 (M8/M9); the set {1.0,1.5,2.0} maximizes average benchmark performance.
  • Input resolution 384x384 = 384x384
    Changing from 352 to 384 improves performance by 1.42% in ablation M4; chosen as the final input size.
  • Loss balancing factor lambda (cosine schedule) = Not numerically specified
    Introduced in Eq. 3 to weight the Uncertainty Awareness Loss; controlled via a cosine schedule and not independently ablated.
  • MGFU group count = 6 = 6
    Hyperparameter of the Multi-Granularity Fusion Unit; fixed without sensitivity analysis.
  • ABSIU attention heads = 4 = 4
    Number of attention groups in the scale-integration unit; fixed without sensitivity analysis.
axioms (4)
  • domain assumption Pretrained PVTv2 ImageNet features transfer to camouflaged-object detection
    The encoder is initialized with ImageNet weights (Section 4.1); if this transfer is not effective, the reported results would change.
  • domain assumption Standard COD benchmark splits are appropriate
    Training on 1,000 CAMO and 3,040 COD10K images and testing on the rest follows prior work; no separate validation split or leakage analysis is reported.
  • domain assumption The UAL loss from ZoomNeXt improves training without introducing hidden benefit
    Borrowed loss term (Eq. 2) is assumed to aid training; its contribution is not isolated in the ablations.
  • domain assumption Published SOTA metrics are reliable and measured under identical protocols
    Table 2 imports all SOTA results from their source papers instead of recomputing them, so comparability rests on the assumption that evaluation procedures match.

pith-pipeline@v1.3.0-alltime-deepseek · 17914 in / 8998 out tokens · 78410 ms · 2026-08-03T21:55:51.384857+00:00 · methodology

0 comments
read the original abstract

Camouflaged object detection is an emerging and challenging computer vision task that requires identifying and segmenting objects that blend seamlessly into their environments due to high similarity in color, texture, and size. This task is further complicated by low-light conditions, partial occlusion, small object size, intricate background patterns, and multiple objects. While many sophisticated methods have been proposed for this task, current methods still struggle to precisely detect camouflaged objects in complex scenarios, especially with small and multiple objects, indicating room for improvement. We propose a Multi-Scale Recursive Network that extracts multi-scale features using a Pyramid Vision Transformer backbone and combines them with specialized Attention-Based Scale Integration Units, thereby enabling selective feature merging. For more precise object detection, our decoder recursively refines features by incorporating Multi-Granularity Fusion Units. A novel recursive-feedback decoding strategy is developed to enhance the model's understanding of global context, thereby helping it overcome the challenges of this task. By jointly leveraging multi-scale learning and recursive feature optimization, our proposed method achieves performance gains, successfully detecting small and multiple camouflaged objects. Our model achieves state-of-the-art results on two benchmark datasets for camouflaged object detection and ranks second on the remaining two. Our code, model weights, and results are available at https://github.com/linaagh98/MSRNet.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

61 extracted references · 8 linked inside Pith

  1. [1]

    In: 9th International Conference on Information Technol- ogy (ICIT’06), pp

    Bhajantri, N.U., Nagabhushan, P.: Camouflage defect identification: a MSRNet: A Multi-Scale Recursive Network for Camouflaged Object Detection19 novel approach. In: 9th International Conference on Information Technol- ogy (ICIT’06), pp. 145–148 (2006). IEEE

    2006

  2. [2]

    Pattern Recognition127, 108644 (2022)

    Zhuge, M., Lu, X., Guo, Y., Cai, Z., Chen, S.: Cubenet: X-shape connec- tion for camouflaged object detection. Pattern Recognition127, 108644 (2022)

    2022

  3. [3]

    Ieri Procedia4, 351–357 (2013)

    Singh, S.K., Dhawale, C.A., Misra, S.: Survey of object detection methods in camouflaged image. Ieri Procedia4, 351–357 (2013)

    2013

  4. [4]

    In: International Conference on Medical Image Computing and Computer- assisted Intervention, pp

    Fan, D.-P., Ji, G.-P., Zhou, T., Chen, G., Fu, H., Shen, J., Shao, L.: Pranet: Parallel reverse attention network for polyp segmentation. In: International Conference on Medical Image Computing and Computer- assisted Intervention, pp. 263–273 (2020). Springer

    2020

  5. [5]

    Machine Intelligence Research19(6), 531–549 (2022)

    Ji, G.-P., Xiao, G., Chou, Y.-C., Fan, D.-P., Zhao, K., Chen, G., Van Gool, L.: Video polyp segmentation: A deep learning perspective. Machine Intelligence Research19(6), 531–549 (2022)

    2022

  6. [6]

    Zhao, X., Zhang, L., Lu, H.: Automatic polyp segmentation via multi- scale subtraction network. In: Medical Image Computing and Computer Assisted Intervention–MICCAI 2021: 24th International Conference, Strasbourg, France, September 27–October 1, 2021, Proceedings, Part I 24, pp. 120–130 (2021). Springer

    2021

  7. [7]

    IEEE transactions on medical imaging39(8), 2626–2637 (2020)

    Fan, D.-P., Zhou, T., Ji, G.-P., Zhou, Y., Chen, G., Fu, H., Shen, J., Shao, L.: Inf-net: Automatic covid-19 lung infection segmentation from ct images. IEEE transactions on medical imaging39(8), 2626–2637 (2020)

    2020

  8. [8]

    Ieee Access7, 45301–45312 (2019)

    Liu, L., Wang, R., Xie, C., Yang, P., Wang, F., Sudirman, S., Liu, W.: Pestnet: An end-to-end deep learning approach for large-scale multi-class pest detection and classification. Ieee Access7, 45301–45312 (2019)

    2019

  9. [9]

    Artificial Intelligence in Agriculture7, 44–57 (2023)

    Rizzo, M., Marcuzzo, M., Zangari, A., Gasparetto, A., Albarelli, A.: Fruit ripeness classification: A survey. Artificial Intelligence in Agriculture7, 44–57 (2023)

    2023

  10. [10]

    IEEE transactions on pattern analysis and machine intelligence44(10), 6024–6042 (2021)

    Fan, D.-P., Ji, G.-P., Cheng, M.-M., Shao, L.: Concealed object detection. IEEE transactions on pattern analysis and machine intelligence44(10), 6024–6042 (2021)

    2021

  11. [11]

    In: Big Data Analysis and Deep Learning Applications: Proceedings of the First International Conference on Big Data Analysis and Deep Learning 1st, pp

    Khaing, M.P., Masayuki, M.: Transparent object detection using con- volutional neural network. In: Big Data Analysis and Deep Learning Applications: Proceedings of the First International Conference on Big Data Analysis and Deep Learning 1st, pp. 86–93 (2019). Springer

    2019

  12. [12]

    IEEE Transactions on Instrumentation and Measurement71, 1–14 (2022)

    Zeng, N., Wu, P., Wang, Z., Li, H., Liu, W., Liu, X.: A small-sized 20MSRNet: A Multi-Scale Recursive Network for Camouflaged Object Detection object detection oriented multi-scale feature fusion approach with appli- cation to defect detection. IEEE Transactions on Instrumentation and Measurement71, 1–14 (2022)

    2022

  13. [13]

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

    Fan, D.-P., Ji, G.-P., Sun, G., Cheng, M.-M., Shen, J., Shao, L.: Camou- flaged object detection. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 2777–2787 (2020)

    2020

  14. [14]

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

    Mei, H., Ji, G.-P., Wei, Z., Yang, X., Wei, X., Fan, D.-P.: Camou- flaged object segmentation with distraction mining. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 8772–8781 (2021)

    2021

  15. [15]

    arXiv preprint arXiv:2105.12555 (2021)

    Sun, Y., Chen, G., Zhou, T., Zhang, Y., Liu, N.: Context-aware cross- level fusion network for camouflaged object detection. arXiv preprint arXiv:2105.12555 (2021)

    Pith/arXiv arXiv 2021

  16. [16]

    In: Pro- ceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp

    Pang, Y., Zhao, X., Xiang, T.-Z., Zhang, L., Lu, H.: Zoom in and out: A mixed-scale triplet network for camouflaged object detection. In: Pro- ceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 2160–2170 (2022)

    2022

  17. [17]

    Optics communications283(21), 4274–4282 (2010)

    Beiderman, Y., Teicher, M., Garcia, J., Mico, V., Zalevsky, Z.: Opti- cal technique for classification, recognition and identification of obscured objects. Optics communications283(21), 4274–4282 (2010)

    2010

  18. [18]

    In: Proceedings Ninth IEEE International Conference on Computer Vision, pp

    Galun, Sharon, Basri, Brandt: Texture segmentation by multiscale aggre- gation of filter responses and shape elements. In: Proceedings Ninth IEEE International Conference on Computer Vision, pp. 716–723 (2003). IEEE

    2003

  19. [19]

    In: 2008 International Conference on Audio, Language and Image Processing, pp

    Guo, H., Dou, Y., Tian, T., Zhou, J., Yu, S.: A robust foreground seg- mentation method by temporal averaging multiple video frames. In: 2008 International Conference on Audio, Language and Image Processing, pp. 878–882 (2008). IEEE

    2008

  20. [20]

    Proceedings of the Royal Society B: Biological Sciences280(1758), 20130064 (2013)

    Hall, J.R., Cuthill, I.C., Baddeley, R., Shohet, A.J., Scott-Samuel, N.E.: Camouflage, detection and identification of moving targets. Proceedings of the Royal Society B: Biological Sciences280(1758), 20130064 (2013)

    2013

  21. [21]

    IEEE transactions on circuits and systems for video technology27(9), 2001–2013 (2016)

    Zhang, X., Zhu, C., Wang, S., Liu, Y., Ye, M.: A bayesian approach to camouflaged moving object detection. IEEE transactions on circuits and systems for video technology27(9), 2001–2013 (2016)

    2001

  22. [22]

    arXiv preprint arXiv:2311.17122 (2023) MSRNet: A Multi-Scale Recursive Network for Camouflaged Object Detection21

    Cheng, S., Ji, G.-P., Qin, P., Fan, D.-P., Zhou, B., Xu, P.: Large model based referring camouflaged object detection. arXiv preprint arXiv:2311.17122 (2023) MSRNet: A Multi-Scale Recursive Network for Camouflaged Object Detection21

    Pith/arXiv arXiv 2023

  23. [23]

    arXiv preprint arXiv:2304.05469 (2023)

    Luo, X.-J., Wang, S., Wu, Z., Sakaridis, C., Cheng, Y., Fan, D.-P., Van Gool, L.: Camdiff: Camouflage image augmentation via diffusion model. arXiv preprint arXiv:2304.05469 (2023)

    Pith/arXiv arXiv 2023

  24. [24]

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

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

    2024

  25. [25]

    arXiv preprint arXiv:2306.07532 (2023)

    Zhang, X., Yin, B., Lin, Z., Hou, Q., Fan, D.-P., Cheng, M.-M.: Referring camouflaged object detection. arXiv preprint arXiv:2306.07532 (2023)

    Pith/arXiv arXiv 2023

  26. [26]

    In: Proceedings of the AAAI Conference on Artificial Intelligence, pp

    Zhu, H., Li, P., Xie, H., Yan, X., Liang, D., Chen, D., Wei, M., Qin, J.: I can find you! boundary-guided separated attention network for cam- ouflaged object detection. In: Proceedings of the AAAI Conference on Artificial Intelligence, pp. 3608–3616 (2022)

    2022

  27. [27]

    Machine Intelligence Research20(1), 92–108 (2023)

    Ji, G.-P., Fan, D.-P., Chou, Y.-C., Dai, D., Liniger, A., Van Gool, L.: Deep gradient learning for efficient camouflaged object detection. Machine Intelligence Research20(1), 92–108 (2023)

    2023

  28. [28]

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

    Yang, F., Zhai, Q., Li, X., Huang, R., Luo, A., Cheng, H., Fan, D.-P.: Uncertainty-guided transformer reasoning for camouflaged object detec- tion. In: Proceedings of the IEEE/CVF International Conference on Computer Vision, pp. 4146–4155 (2021)

    2021

  29. [29]

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

    Jia, Q., Yao, S., Liu, Y., Fan, X., Liu, R., Luo, Z.: Segment, magnify and reiterate: Detecting camouflaged objects the hard way. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 4713–4722 (2022)

    2022

  30. [30]

    In: Proceedings of the 30th ACM International Conference on Multimedia, pp

    Zhang, M., Xu, S., Piao, Y., Shi, D., Lin, S., Lu, H.: Preynet: Preying on camouflaged objects. In: Proceedings of the 30th ACM International Conference on Multimedia, pp. 5323–5332 (2022)

    2022

  31. [31]

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

    Lv, Y., Zhang, J., Dai, Y., Li, A., Liu, B., Barnes, N., Fan, D.-P.: Simultaneously localize, segment and rank the camouflaged objects. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 11591–11601 (2021)

    2021

  32. [32]

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

    Zhai, Q., Li, X., Yang, F., Chen, C., Cheng, H., Fan, D.-P.: Mutual graph learning for camouflaged object detection. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 12997–13007 (2021)

    2021

  33. [33]

    In: Proceedings of 22MSRNet: A Multi-Scale Recursive Network for Camouflaged Object Detection the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp

    Li, A., Zhang, J., Lv, Y., Liu, B., Zhang, T., Dai, Y.: Uncertainty-aware joint salient object and camouflaged object detection. In: Proceedings of 22MSRNet: A Multi-Scale Recursive Network for Camouflaged Object Detection the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 10071–10081 (2021)

    2021

  34. [34]

    arXiv preprint arXiv:2207.00794 (2022)

    Sun, Y., Wang, S., Chen, C., Xiang, T.-Z.: Boundary-guided camouflaged object detection. arXiv preprint arXiv:2207.00794 (2022)

    Pith/arXiv arXiv 2022

  35. [35]

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

    He, C., Li, K., Zhang, Y., Tang, L., Zhang, Y., Guo, Z., Li, X.: Camouflaged object detection with feature decomposition and edge recon- struction. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 22046–22055 (2023)

    2023

  36. [36]

    arXiv preprint arXiv:2010.11929 (2020)

    Dosovitskiy, A.: An image is worth 16x16 words: Transformers for image recognition at scale. arXiv preprint arXiv:2010.11929 (2020)

    Pith/arXiv arXiv 2010

  37. [37]

    In: International Conference on Machine Learning, pp

    Touvron, H., Cord, M., Douze, M., Massa, F., Sablayrolles, A., J´ egou, H.: Training data-efficient image transformers & distillation through atten- tion. In: International Conference on Machine Learning, pp. 10347–10357 (2021). PMLR

    2021

  38. [38]

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

    Wang, W., Xie, E., Li, X., Fan, D.-P., Song, K., Liang, D., Lu, T., Luo, P., Shao, L.: Pyramid vision transformer: A versatile backbone for dense prediction without convolutions. In: Proceedings of the IEEE/CVF International Conference on Computer Vision, pp. 568–578 (2021)

    2021

  39. [39]

    Advances in neural information processing systems34, 18590–18602 (2021)

    Jiang, Z.-H., Hou, Q., Yuan, L., Zhou, D., Shi, Y., Jin, X., Wang, A., Feng, J.: All tokens matter: Token labeling for training better vision transform- ers. Advances in neural information processing systems34, 18590–18602 (2021)

    2021

  40. [40]

    Advances in neural information processing systems34, 12077–12090 (2021)

    Xie, E., Wang, W., Yu, Z., Anandkumar, A., Alvarez, J.M., Luo, P.: Segformer: Simple and efficient design for semantic segmentation with transformers. Advances in neural information processing systems34, 12077–12090 (2021)

    2021

  41. [41]

    In: European Confer- ence on Computer Vision, pp

    Carion, N., Massa, F., Synnaeve, G., Usunier, N., Kirillov, A., Zagoruyko, S.: End-to-end object detection with transformers. In: European Confer- ence on Computer Vision, pp. 213–229 (2020). Springer

    2020

  42. [42]

    In: Proceedings of the AAAI Conference on Artificial Intelligence, pp

    Gu, Y., Wang, L., Wang, Z., Liu, Y., Cheng, M.-M., Lu, S.-P.: Pyramid constrained self-attention network for fast video salient object detection. In: Proceedings of the AAAI Conference on Artificial Intelligence, pp. 10869–10876 (2020)

    2020

  43. [43]

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

    Liu, N., Zhang, N., Wan, K., Shao, L., Han, J.: Visual saliency trans- former. In: Proceedings of the IEEE/CVF International Conference on Computer Vision, pp. 4722–4732 (2021) MSRNet: A Multi-Scale Recursive Network for Camouflaged Object Detection23

    2021

  44. [44]

    IEEE Transactions on Pattern Analysis and Machine Intelligence45(3), 3738–3752 (2022)

    Zhuge, M., Fan, D.-P., Liu, N., Zhang, D., Xu, D., Shao, L.: Salient object detection via integrity learning. IEEE Transactions on Pattern Analysis and Machine Intelligence45(3), 3738–3752 (2022)

    2022

  45. [45]

    IEEE Transactions on Circuits and Systems for Video Technology 33(9), 4934–4947 (2023)

    Liu, Y., Li, H., Cheng, J., Chen, X.: Mscaf-net: A general framework for camouflaged object detection via learning multi-scale context-aware fea- tures. IEEE Transactions on Circuits and Systems for Video Technology 33(9), 4934–4947 (2023)

    2023

  46. [46]

    IEEE Transactions on Circuits and Systems for Video Technology33(10), 5444–5457 (2023)

    Xing, H., Gao, S., Wang, Y., Wei, X., Tang, H., Zhang, W.: Go closer to see better: Camouflaged object detection via object area amplification and figure-ground conversion. IEEE Transactions on Circuits and Systems for Video Technology33(10), 5444–5457 (2023)

    2023

  47. [47]

    In: Proceedings of the AAAI Conference on Artificial Intelligence, pp

    Hu, X., Wang, S., Qin, X., Dai, H., Ren, W., Luo, D., Tai, Y., Shao, L.: High-resolution iterative feedback network for camouflaged object detec- tion. In: Proceedings of the AAAI Conference on Artificial Intelligence, pp. 881–889 (2023)

    2023

  48. [48]

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

    Huang, Z., Dai, H., Xiang, T.-Z., Wang, S., Chen, H.-X., Qin, J., Xiong, H.: Feature shrinkage pyramid for camouflaged object detection with transformers. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 5557–5566 (2023)

    2023

  49. [49]

    The Visual Computer39(10), 4593–4607 (2023)

    Zhang, Q., Ge, Y., Zhang, C., Bi, H.: Tprnet: camouflaged object detec- tion via transformer-induced progressive refinement network. The Visual Computer39(10), 4593–4607 (2023)

    2023

  50. [50]

    In: 2022 26th International Conference on Pattern Recognition (ICPR), pp

    Liu, Z., Zhang, Z., Tan, Y., Wu, W.: Boosting camouflaged object detec- tion with dual-task interactive transformer. In: 2022 26th International Conference on Pattern Recognition (ICPR), pp. 140–146 (2022). IEEE

    2022

  51. [51]

    IEEE Transactions on Pattern Analysis and Machine Intelligence (2024)

    Yin, B., Zhang, X., Fan, D.-P., Jiao, S., Cheng, M.-M., Van Gool, L., Hou, Q.: Camoformer: Masked separable attention for camouflaged object detection. IEEE Transactions on Pattern Analysis and Machine Intelligence (2024)

    2024

  52. [52]

    IEEE Transactions on Pattern Analysis and Machine Intelligence (2024)

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

    2024

  53. [53]

    Computational Visual Media8(3), 415–424 (2022)

    Wang, W., Xie, E., Li, X., Fan, D.-P., Song, K., Liang, D., Lu, T., Luo, P., Shao, L.: Pvt v2: Improved baselines with pyramid vision transformer. Computational Visual Media8(3), 415–424 (2022)

    2022

  54. [54]

    Computer vision and image 24MSRNet: A Multi-Scale Recursive Network for Camouflaged Object Detection understanding184, 45–56 (2019)

    Le, T.-N., Nguyen, T.V., Nie, Z., Tran, M.-T., Sugimoto, A.: Anabranch network for camouflaged object segmentation. Computer vision and image 24MSRNet: A Multi-Scale Recursive Network for Camouflaged Object Detection understanding184, 45–56 (2019)

    2019

  55. [55]

    Unpublished manuscript2(6), 7 (2018)

    Przemys law, S., Hassan, A., Jakub, B., Tomasz, D., Adam, K., Kozie l, P.: Animal camouflage analysis: Chameleon database. Unpublished manuscript2(6), 7 (2018)

    2018

  56. [56]

    In: Proceedings of the IEEE International Conference on Computer Vision, pp

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

    2017

  57. [57]

    248–255 (2014)

    Margolin, R., Zelnik-Manor, L., Tal, A.: How to evaluate foreground maps? In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pp. 248–255 (2014)

    2014

  58. [58]

    arXiv preprint arXiv:1805.10421 (2018)

    Fan, D.-P., Gong, C., Cao, Y., Ren, B., Cheng, M.-M., Borji, A.: Enhanced-alignment measure for binary foreground map evaluation. arXiv preprint arXiv:1805.10421 (2018)

    Pith/arXiv arXiv 2018

  59. [59]

    In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pp

    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

  60. [60]

    IEEE transactions on pattern analysis and machine intelligence43(2), 652–662 (2019)

    Gao, S.-H., Cheng, M.-M., Zhao, K., Zhang, X.-Y., Yang, M.-H., Torr, P.: Res2net: A new multi-scale backbone architecture. IEEE transactions on pattern analysis and machine intelligence43(2), 652–662 (2019)

    2019

  61. [61]

    arXiv preprint arXiv:1905.11946 (2019)

    Tan, M.: Efficientnet: Rethinking model scaling for convolutional neural networks. arXiv preprint arXiv:1905.11946 (2019)

    Pith/arXiv arXiv 1905