Pith. sign in

REVIEW 5 major objections 6 minor 77 references

FOAM: A General Frequency-Optimized Anti-Overlapping Framework for Overlapping Object Perception

T0 review · 5 major / 6 minor · reviewed 2026-08-15 · deepseek-v4-flash

Pith's one-line read FOAM boosts DINO by 2.9 AP on overlapping X-ray detection

desk verdict A plausible incremental architecture whose headline AP gains are inflated by test-set hyperparameter selection, but the core idea is worth a serious look after fixes. read the letter →

arxiv 2506.13501 v1 pith:ZWWSRMOJ submitted 2025-06-16 cs.CV

classification cs.CV
keywords overlappingobjectperceptionfrequencydomainlearningmagnitudespectrumX-rayprohibiteditemdetectioninstancesegmentationpneumoniaconsistencyloss
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

The paper tries to establish that overlapping-object perception—detecting or segmenting foreground objects whose contours and textures are partly buried in background clutter—is materially improved by learning from the Fourier magnitude spectrum, not just from spatial features. It claims that the degradation from overlap shows up clearly in the magnitude spectrum, and that a plugin framework called FOAM lets a model re-weight frequency bands to recover texture and contour cues. FOAM's Frequency Spatial Transformer Block (FSTB) combines frequency and spatial self-attention, while its Hierarchical De-Corrupting (HDC) mechanism trains a corrupted-image branch to match the clean branch, teaching the network to suppress background response. If the paper is right, adding FOAM to an existing detector or segmenter is a general way to raise accuracy on security-screening and medical X-ray tasks without changing the task-specific head.

What carries the argument

The Frequency Spatial Transformer Block (FSTB) is the core module: it runs a channel self-attention over spatial features (SDCA), a spatial self-attention over the FFT magnitude spectrum (FDBA) that reweights frequency bands while leaving the phase untouched, and a feed-forward network (FSFN) that fuses the two domains. Cascaded FSTBs form the base branch, which runs at inference. The Hierarchical De-Corrupting (HDC) mechanism is the training-time companion: it applies Gaussian blur, downsampling, or noise to create a corrupted image, runs the same FSTBs on it through shared weights, and uses a KL-based consistent loss to align corrupted features with cleaner base-branch features, with an MSE variant as an alternative. The stated rationale is that suppressing background response increases foreground-to-background contrast at contours, expressed as Theorem I's inequality.

What would settle it

Measure foreground-background contrast in the feature maps of DINO, DINO+FSTB, and FSTB+HDC under identical training budgets. If box AP rises with HDC but average contour contrast in the feature maps does not increase, then the claimed suppression mechanism is not what drives the gain. Alternatively, replace the corrupted-image branch with an unrelated randomly blurred image under the same loss; equal gains would indicate the benefit is generic regularization rather than learned de-corruption.

Watch

Extended reading notes

Core claim

The central discovery is that the magnitude spectrum of an image is a reliable signal for overlapping-object perception: when contours and textures are degraded by overlap, the magnitude spectrum changes in characteristic ways, while the phase spectrum and spatial appearance stay comparatively stable. FOAM exploits this by decoupling magnitude from phase: FDBA applies self-attention across frequency bands of the magnitude spectrum to reweight texture-carrying bands, the phase is preserved, and an inverse Fourier transform returns a corrected feature map that is fused with spatial features. The HDC mechanism then creates a second branch on a deliberately corrupted image and aligns its features with the clean branch via a KL-divergence consistent loss, so the FSTB learns to suppress background features and sharpen foreground contours. The paper reports that this raises DINO from 64.3 to 67.2 box AP on PIXray-det with ResNet-50, that F-DINO with Swin-L reaches 73.7 AP and 81.7 mAP on OPIXray, and that the same recipe improves instance segmentation and pneumonia detection.

Load-bearing premise

The load-bearing premise is that training a deliberately corrupted branch to match the clean branch suppresses background responses and sharpens foreground contours; if that premise fails, HDC contributes nothing beyond the FSTB.

Editorial extensions

If this is right

  • Plugging FOAM into a detector or segmenter improves accuracy without changing the task-specific head: F-DINO, F-Mask-R-CNN, and F-Cascade-R-CNN all beat their bases.
  • FOAM generalizes across CNN and Transformer backbones (ResNet-50, ResNeXt-101, Swin-L) and across detection and instance segmentation architectures including Deformable-DETR, DINO, RT-DETR, Mask2Former, CondInst, Cascade-Mask-R-CNN, and Mask-R-CNN.
  • Because the corruption branch is used only at training time, inference cost is just the added FSTBs; the paper reports about 1.4M extra parameters and 4.0 GFLOPs for the N=1 DINO configuration.
  • The consistent loss choice matters: KL-divergence (Type I) gives higher and more stable AP than MSE (Type II), and Gaussian blur is the most effective corruption strategy, with kernel size 3 and sigma 5.
  • Aligning only higher-level multi-scale layers yields the best results; including low-level layers hurts, since low-level features carry noise.

Reading between the lines

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

  • Because HDC needs no overlap annotations, FOAM should transfer to any heavily occluded detection domain, such as crowd counting, autonomous driving, or camouflaged objects, provided the magnitude-spectrum degradation assumption holds.
  • The paper leaves open whether reweighting the magnitude spectrum is the active ingredient or whether any global frequency perturbation would regularize equally well; a controlled swap of phase versus magnitude corruption would settle this.
  • The HDC mechanism resembles self-supervised denoising, so a natural extension is to use the corruption branch for test-time adaptation on unlabeled degraded images from a target hospital or scanner.
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 / 6 minor

Summary. The paper proposes FOAM, a plug-in framework for overlapping-object perception. It inserts Frequency Spatial Transformer Blocks (FSTBs) after the backbone to combine spatial- and frequency-domain cues, and adds a training-only Hierarchical De-Corrupting (HDC) branch that aligns features from a corrupted image to the base-branch features via a consistency loss. FOAM is evaluated by adding it to DETR-based, CNN-based, and segmentation detectors on PIXray-det, OPIXray, PIXray-seg, and RLP, reporting gains over baselines with the largest gain being 2.9 AP for F-DINO over DINO on PIXray-det. The paper includes extensive ablations of corruption strategies, loss types, target layers, and attention-map shapes, together with qualitative visualizations of features, sampling points, and predictions.

Significance. If confirmed, FOAM would be a practically useful plug-in: it is architecture-agnostic, improves several modern detectors, and addresses a real need in security X-ray and medical imaging. The breadth of the experiments (four datasets, three tasks, multiple backbones) is a strength, as is the coverage of design choices in the ablations. However, the current evidence is not yet convincing: the headline configuration is selected on the test set, several reported gains are within typical run-to-run variation, the two closest frequency-domain competitors are missing from the comparisons, and the mechanism is not precisely defined in the equations. The contribution is incremental and modular rather than a new task or paradigm; with corrected evaluation and added comparisons it could become a solid applied contribution.

major comments (5)
  1. [Sec. IV-E, Table VII, Fig. 5] The empirical protocol in Sec. IV-E selects the final model by maximizing box AP on the PIXray-det test set. Table VII evaluates DU, GN, and GB at several severity levels each and adopts the single best setting (GB, ks=3, sigma=5), and Fig. 5 then chooses the consistent-loss type, its coefficient, and the learning rate using the same test set. Because the reported F-DINO result of 67.2 AP is the maximum over this grid, the headline 2.9 AP gain over DINO is not an unbiased estimate of FOAM's effect. Please switch to a validation-based selection procedure (or nested cross-validation) and report only the test-set result of the configuration chosen on the validation set.
  2. [Table I and Table V] Several of the headline improvements are very small and are reported without variance. For example, Table I reports F-RT-DETR at 61.7 vs. RT-DETR at 61.4 (+0.3 AP), F-DINO(Swin-L) at 73.7 vs. DINO(Swin-L) at 73.3 (+0.4 AP), and F-Mask-R-CNN(X-101) at 65.6 vs. 65.2 (+0.4 AP); Table V reports F-Cascade-R-CNN at 20.2 vs. Cascade-R-CNN at 19.7 (+0.5 AP). With single runs and no error bars, these differences are consistent with training noise under the 12-epoch protocol. Please report means and standard deviations over multiple seeds, or otherwise demonstrate that the differences are statistically stable.
  3. [Sec. I, Sec. II-A, Tables II-IV] The manuscript motivates FOAM against FAPID [12] and FDTNet [14], stating that their fixed high-pass and SRM filters are limiting, yet neither method appears in the quantitative comparisons in Tables II-IV. Because they are the closest prior frequency-domain approaches for prohibited-item detection, their omission prevents the reader from assessing the central improvement claim over the specialized state of the art. Please include them under the same training and evaluation protocol as the other baselines.
  4. [Sec. III-B2, Eqs. (8)-(11), Theorem I] The HDC loss is defined inconsistently. Sec. III-B2 states that the loss supervises F^{n+1}_C with F^n_O, but Eq. (8) compares \hat F^{n-1}_{O,l} with \hat F^n_{C,l}, and Eq. (9) normalizes by \sum_i \exp(F^n_{O,l}(i)) while the numerator is \exp(F^{n-1}_{O,l}(i)). Additionally, Theorem I analyzes a scalar additive response f+b and shows that reducing b increases the contrast ratio; it does not establish that minimizing the KL loss in Eq. (8) suppresses background response in deep feature maps. Please correct the indexing and either prove the claimed suppression under explicit assumptions or label Theorem I as intuition rather than a formal justification.
  5. [Table I vs. Table II; Table VII vs. Table VIII] Numbers across tables are not mutually consistent. Table I reports RT-DETR (ResNet-50) at 61.4 APbox, while Table II reports 62.3 for the same detector and backbone; Table VIII's caption says the experiment uses Gaussian Blurring with kernel size 5 and sigma 5, but its best row (l3,l4) gives 67.2, the same value reported for the final model that Table VII selects with ks=3. The reader cannot tell which corruption setting produced which result. Please align the baselines and state the exact settings used in every ablation.
minor comments (6)
  1. [Eq. (29)] Eq. (29) has unbalanced parentheses and an ambiguous norm placement; please rewrite it so that the operator precedence is clear.
  2. [Table III caption] The caption lists four category names for five columns (FO, ST, SC, UT, MU) and omits the definition of SC; please include all five category definitions.
  3. [Table VII caption] The caption states that 'KS means the kernel size of Gaussian Noise'; it should say Gaussian Blurring.
  4. [Fig. 5] The boxplot is referenced but the exact grid values for the learning rate and loss coefficient are not reported in the text or caption; please list the tested values.
  5. [Sec. IV-A, Table VI] The text says 'FPS decreased by 16 frames'; FPS is already a rate, so please rephrase this as a decrease of 16 frames per second.
  6. [Sec. III-B2] The statement that 'the exact form of the consistent loss is not crucial' is hard to reconcile with the ablation in Sec. IV-E3 showing a preference for the Type I loss; please clarify or qualify the statement.

Circularity Check

1 steps flagged · score 6.0 of 10

Test-set hyperparameter selection makes F-DINO's 67.2 AP the argmax over the searched grid; otherwise no load-bearing self-citation or definitional circularity.

  1. fitted input called prediction [Sec. IV-E2 (Table VII), Sec. IV-E3 (Fig. 5), and Sec. IV-D1 (Tables I-II)]
    "Furthermore, when ks = 3 and the Gaussian Blurring sigma is set to 5, the model achieved its highest box AP of 67.2%."

    The HDC hyperparameters (corruption type, severity, loss type, loss coefficient, learning rate) are selected by maximizing box AP on the PIXray-det test set: Table VII sweeps GB/DU/GN and adopts the single configuration with the highest AP, while Fig. 5 tunes loss type and coefficients on the same test set. The F-DINO AP of 67.2 reported as the headline result is exactly this test-set-best value, so the claimed 2.9-point gain over DINO is the maximum of the searched grid rather than the independent predictive performance of a pre-specified method. The evaluation metric is thereby used as the fit criterion, making the central PIXray-det improvement statistically forced by construction.

full rationale

The paper's theoretical support (Theorem I, Eq. 12) is only a tautological inequality about linear additive responses and does not constrain the FSTB/HDC design, so it is weak but not circular. The self-citations (AO-DETR, MMCL, CSPCL, PIXray) appear as baselines and benchmark data, not as load-bearing derivation premises, so no self-citation chain forces the result. The one genuine circularity is in the validation protocol: corruption strategy, severity, loss type, coefficient, and learning rate are chosen on the PIXray-det test set, and the resulting 67.2 AP is then reported as F-DINO's performance. This makes the primary headline gain a selected maximum rather than an unbiased prediction. Because the same settings are reused without per-dataset grid search on OPIXray, RLP, and PIXray-seg, the circularity is partial, but the main PIXray-det quantitative claim is statistically forced by the selection criterion.

Assumptions & free parameters 7 free parameters · 4 assumptions · 5 invented entities

The framework introduces several new components (FSTB, FDBA, SDCA, FSFN, HDC) and several hyperparameters chosen from test-set ablations. The central empirical claim is supported only by internal evaluations on the same datasets, with no external predictions; the theoretical motivation (Theorem I) is a trivial inequality that does not justify the mechanisms.

free parameters (7)
  • Gaussian blur sigma for HDC corruption = 5
    Selected by test-set search (Table VII: GB ks=3 sigma=5 gives 67.2 AP; sigma=10 gives 66.5).
  • Gaussian blur kernel size = 3
    Test-set ablation (Table VII) shows ks=3 beats ks=5 for all sigmas.
  • Number of cascaded FSTBs (N) = 1
    From Table VI, N=1 gives 67.2 AP vs N=2 (67.3) and N=3 (67.7); chosen for efficiency, not for accuracy.
  • Target layer set L for consistent loss = {3,4}
    From Table VIII, L={3,4} gives 67.2 AP, while L={4} gives 67.1 and all layers gives 65.2.
  • Consistent loss variant = Type I (KL divergence)
    From Fig. 5, Type I has higher mean and lower variance than Type II.
  • Corruption strategy = Gaussian Blurring
    From Table VII, GB gives the highest AP (67.2) among GB, DU, and GN strategies.
  • Consistency loss weight = not reported in final details
    A coefficient is swept in Fig. 5, but the final value used in the main experiments is not stated in the implementation details.
assumptions (4)
  • domain assumption The FFT magnitude spectrum of deep features reflects texture and contour degradation
    Sec. III-A and Fig. 2 show this for raw images under synthetic corruptions; the paper assumes it transfers to deep feature maps.
  • domain assumption Corrupting the input simulates more severe overlap and provides a useful training signal
    Sec. III-B2, HDC corrupts the image to simulate severe overlap; it is assumed this teaches robustness to background clutter.
  • ad hoc to paper Reducing background response increases foreground contour contrast (linear model f+b)
    Theorem I, Eq. 12; assumes an additive linear response of a network to foreground and background, which is not how deep networks behave.
  • standard math Standard properties of FFT/IFFT and KL divergence
    Used in Sec. III-A and Eq. 8 without proof, relying on textbook Fourier analysis.
invented entities (5)
  • Frequency Spatial Transformer Block (FSTB)
    purpose: Adds frequency and spatial feature extraction to the backbone
    New module; its benefit is shown only via internal ablations, with no external falsifiable prediction.
  • Hierarchical De-Corrupting (HDC) mechanism
    purpose: Training-time consistency between base and corrupted branches to suppress background response
    New training mechanism; effectiveness is empirical only, with no independent handle.
  • Frequency Domain Bands Self-Attention (FDBA)
    purpose: Attention on the magnitude spectrum to refine frequency bands
    Internal submodule of FSTB; no independent evidence outside this paper.
  • Spatial Domain Channel Self-Attention (SDCA)
    purpose: Channel attention over spatial features
    Internal submodule of FSTB; no independent evidence outside this paper.
  • Frequency Spatial Feed-forward Network (FSFN)
    purpose: Integrates frequency and spatial features
    Internal submodule of FSTB; no independent evidence outside this paper.

how reviews work

0 comments
Cite this review

Pith. "Pith review of FOAM: A General Frequency-Optimized Anti-Overlapping Framework for Overlapping Object Perception." pith.science (2026). https://pith.science/paper/ZWWSRMOJ

@misc{pith2026250613501,
  author       = {Pith},
  title        = {Pith review of: FOAM: A General Frequency-Optimized Anti-Overlapping Framework for Overlapping Object Perception},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/ZWWSRMOJ}},
  note         = {Machine review of arXiv:2506.13501}
}
read the original abstract

Overlapping object perception aims to decouple the randomly overlapping foreground-background features, extracting foreground features while suppressing background features, which holds significant application value in fields such as security screening and medical auxiliary diagnosis. Despite some research efforts to tackle the challenge of overlapping object perception, most solutions are confined to the spatial domain. Through frequency domain analysis, we observe that the degradation of contours and textures due to the overlapping phenomenon can be intuitively reflected in the magnitude spectrum. Based on this observation, we propose a general Frequency-Optimized Anti-Overlapping Framework (FOAM) to assist the model in extracting more texture and contour information, thereby enhancing the ability for anti-overlapping object perception. Specifically, we design the Frequency Spatial Transformer Block (FSTB), which can simultaneously extract features from both the frequency and spatial domains, helping the network capture more texture features from the foreground. In addition, we introduce the Hierarchical De-Corrupting (HDC) mechanism, which aligns adjacent features in the separately constructed base branch and corruption branch using a specially designed consistent loss during the training phase. This mechanism suppresses the response to irrelevant background features of FSTBs, thereby improving the perception of foreground contour. We conduct extensive experiments to validate the effectiveness and generalization of the proposed FOAM, which further improves the accuracy of state-of-the-art models on four datasets, specifically for the three overlapping object perception tasks: Prohibited Item Detection, Prohibited Item Segmentation, and Pneumonia Detection. The code will be open source once the paper is accepted.

Figures

Figures reproduced from arXiv: 2506.13501 by the authors.

Figure 1
Figure 1. The overlapping phenomenon causes the contours and textures of [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. X-ray images analysis in spatial and frequency domains. GB, DU, [PITH_FULL_IMAGE:figures/full_fig_p002_2.png] view at source ↗
Figure 3
Figure 3. Overall architecture of our proposed FOAM. The core module is FSTB, which is designed for joint learning in both the frequency and spatial [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figures from the paper (5 more)
Figure 4
Figure 4. Figure 4: The detailed flowchart of the proposed core building module FSTB, consisting of three components: Spatial Domain Channel Self-Attention (SDCA), [PITH_FULL_IMAGE:figures/full_fig_p007_4.png]
Figure 5
Figure 5. Figure 5: Comparison of Type I consistent Loss and Type II consistent Loss. The [PITH_FULL_IMAGE:figures/full_fig_p012_5.png]
Figure 6
Figure 6. Figure 6: Visualization of feature maps of “DINO”, “DINO+FSTB”, and [PITH_FULL_IMAGE:figures/full_fig_p013_6.png]
Figure 8
Figure 8. Figure 8: The scatter plot and Kernel Density Estimation (KDE) joint distri [PITH_FULL_IMAGE:figures/full_fig_p013_8.png]
Figure 9
Figure 9. Figure 9: Prediction results on PIXray-det, OPIXray, RLP, PIXray-seg datasets. We select state-of-the-art models from four datasets across three tasks (row two) [PITH_FULL_IMAGE:figures/full_fig_p014_9.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

77 extracted references · 70 canonical work pages

  1. [12]

    Feature-aware prohibited items detection for x-ray images,

    H. Liao, B. Huang, and H. Gao, “Feature-aware prohibited items detection for x-ray images,” in 2023 IEEE International Conference on Image Processing (ICIP) . IEEE, 2023, pp. 1040–1044. 2, 3, 6

  2. [14]

    Fdtnet: Enhanc- ing frequency-aware representation for prohibited object detection from x-ray images via dual-stream transformers,

    Z. Zhu, Y . Zhu, H. Wang, N. Wang, J. Ye, and X. Ling, “Fdtnet: Enhanc- ing frequency-aware representation for prohibited object detection from x-ray images via dual-stream transformers,” Engineering Applications of Artificial Intelligence , vol. 133, p. 108076, 2024. 2, 3, 6

  3. [1]

    Automated segmentation of prohibited items in x-ray baggage images using dense de-overlap attention snake,

    B. Ma, T. Jia, M. Su, X. Jia, D. Chen, and Y . Zhang, “Automated segmentation of prohibited items in x-ray baggage images using dense de-overlap attention snake,” IEEE Transactions on Multimedia, 2022. 1, 2, 4, 8, 9, 10, 11, 13

  4. [2]

    Occluded prohibited items detection: An x-ray security inspection benchmark and de-occlusion attention module,

    Y . Wei, R. Tao, Z. Wu, Y . Ma, L. Zhang, and X. Liu, “Occluded prohibited items detection: An x-ray security inspection benchmark and de-occlusion attention module,” in Proceedings of the 28th ACM International Conference on Multimedia , 2020, pp. 138–146. 1, 2, 4, 8, 9, 10

  5. [3]

    Augmenting the national institutes of health chest radiograph dataset with expert annotations of possible pneumonia,

    G. Shih, C. C. Wu, S. S. Halabi, M. D. Kohli, L. M. Prevedello, T. S. Cook, A. Sharma, J. K. Amorosa, V . Arteaga, M. Galperin-Aizenberg et al. , “Augmenting the national institutes of health chest radiograph dataset with expert annotations of possible pneumonia,” Radiology: Artificial Intelligence, vol. 1, no. 1, p. e180041, 2019. 1, 8

  6. [4]

    Chest x- ray interpretation: detecting devices and device-related complications,

    M. Gambato, N. Scotti, G. Borsari, J. Zambon Bertoja, J.-D. Gabrieli, A. De Cassai, G. Cester, P. Navalesi, E. Quaia, and F. Causin, “Chest x- ray interpretation: detecting devices and device-related complications,” Diagnostics, vol. 13, no. 4, p. 599, 2023. 1, 15

  7. [5]

    Gadet: A geometry-aware x-ray prohibited items detector,

    M. Li, B. Ma, H. Wang, D. Chen, and T. Jia, “Gadet: A geometry-aware x-ray prohibited items detector,” IEEE Sensors Journal , vol. 24, no. 2, pp. 1665–1678, 2024. 2, 4

  8. [6]

    Ao-detr: Anti-overlapping detr for x-ray prohibited items detection,

    M. Li, T. Jia, H. Wang, B. Ma, H. Lu, S. Lin, D. Cai, and D. Chen, “Ao-detr: Anti-overlapping detr for x-ray prohibited items detection,” IEEE Transactions on Neural Networks and Learning Systems , pp. 1– 15, 2024. 2, 4, 9, 10

Show all 77 references
  1. [7]

    Detecting prohibited objects with physical size constraint from cluttered x-ray baggage images,

    A. Chang, Y . Zhang, S. Zhang, L. Zhong, and L. Zhang, “Detecting prohibited objects with physical size constraint from cluttered x-ray baggage images,” Knowledge-Based Systems, vol. 237, p. 107916, 2022. 2, 4

  2. [8]

    Detecting overlapped objects in x-ray security imagery by a label-aware mechanism,

    C. Zhao, L. Zhu, S. Dou, W. Deng, and L. Wang, “Detecting overlapped objects in x-ray security imagery by a label-aware mechanism,” IEEE Transactions on Information Forensics and Security , vol. 17, pp. 998– 1009, 2022. 2, 4

  3. [9]

    Sixray: A large-scale security inspection x-ray benchmark for prohibited item discovery in overlapping images,

    C. Miao, L. Xie, F. Wan, C. Su, H. Liu, J. Jiao, and Q. Ye, “Sixray: A large-scale security inspection x-ray benchmark for prohibited item discovery in overlapping images,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2019, pp. 2119–

  4. [10]

    Pid-yolox: An x-ray prohibited items detector based on yolox,

    M. Li, B. Ma, H. Wang, Y . Li, D. Chen, and T. Jia, “Pid-yolox: An x-ray prohibited items detector based on yolox,” in 2023 IEEE 13th International Conference on CYBER Technology in Automation, Control, and Intelligent Systems (CYBER) . IEEE, 2023, pp. 413–418. 2, 3

  5. [11]

    Edge perception camouflaged object detection under frequency domain reconstruction,

    Z. Liu, X. Deng, P. Jiang, C. Lv, G. Min, and X. Wang, “Edge perception camouflaged object detection under frequency domain reconstruction,” IEEE Transactions on Circuits and Systems for Video Technology, 2024. 2

  6. [13]

    The fast fourier trans- form and its applications,

    J. W. Cooley, P. A. W. Lewis, and P. D. Welch, “The fast fourier trans- form and its applications,” IEEE Transactions on Education , vol. 12, no. 1, pp. 27–34, 1969. 2, 3, 4

  7. [15]

    Rich models for steganalysis of digital images,

    J. Fridrich and J. Kodovsky, “Rich models for steganalysis of digital images,” IEEE Transactions on information Forensics and Security , vol. 7, no. 3, pp. 868–882, 2012. 2, 3

  8. [16]

    Frequency separation for real- world super-resolution,

    M. Fritsche, S. Gu, and R. Timofte, “Frequency separation for real- world super-resolution,” in 2019 IEEE/CVF International Conference on Computer Vision Workshop (ICCVW). IEEE, 2019, pp. 3599–3608. 3

  9. [17]

    A frequency domain approach to registration of aliased images with application to super- resolution,

    P. Vandewalle, S. S ¨usstrunk, and M. Vetterli, “A frequency domain approach to registration of aliased images with application to super- resolution,” EURASIP journal on advances in signal processing , vol. 2006, pp. 1–14, 2006. 3

  10. [18]

    Performance analysis of image compression using wavelets,

    S. Grgic, M. Grgic, and B. Zovko-Cihlar, “Performance analysis of image compression using wavelets,” IEEE Transactions on industrial electronics, vol. 48, no. 3, pp. 682–695, 2001. 3

  11. [19]

    Space-frequency quantization for image compression with directionlets,

    V . Velisavljevic, B. Beferull-Lozano, and M. Vetterli, “Space-frequency quantization for image compression with directionlets,” IEEE Transac- tions on Image Processing , vol. 16, no. 7, pp. 1761–1773, 2007. 3

  12. [20]

    Learning in the frequency domain,

    K. Xu, M. Qin, F. Sun, Y . Wang, Y .-K. Chen, and F. Ren, “Learning in the frequency domain,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , June 2020. 3

  13. [21]

    A general spatial-frequency learning framework for multimodal image fusion,

    M. Zhou, J. Huang, K. Yan, D. Hong, X. Jia, J. Chanussot, and C. Li, “A general spatial-frequency learning framework for multimodal image fusion,” IEEE Transactions on Pattern Analysis and Machine Intelligence, 2024. 3

  14. [22]

    Frequency-spatial entanglement learning for camouflaged object detection,

    Y . Sun, C. Xu, J. Yang, H. Xuan, and L. Luo, “Frequency-spatial entanglement learning for camouflaged object detection,” in European Conference on Computer Vision . Springer, 2025, pp. 343–360. 3

  15. [23]

    An enhanced vi- sion transformer with wavelet position embedding for histopathological image classification,

    M. Ding, A. Qu, H. Zhong, Z. Lai, S. Xiao, and P. He, “An enhanced vi- sion transformer with wavelet position embedding for histopathological image classification,” Pattern Recognition, vol. 140, p. 109532, 2023. 3 JOURNAL OF LATEX CLASS FILES, VOL. 14, NO. 8, AUGUST 2021 16

  16. [24]

    Edge perception camouflaged object detection under frequency domain reconstruction,

    Z. Liu, X. Deng, P. Jiang, C. Lv, G. Min, and X. Wang, “Edge perception camouflaged object detection under frequency domain reconstruction,” IEEE Transactions on Circuits and Systems for Video Technology, 2024. 3

  17. [25]

    Spatial-frequency domain information integration for pan-sharpening,

    M. Zhou, J. Huang, K. Yan, H. Yu, X. Fu, A. Liu, X. Wei, and F. Zhao, “Spatial-frequency domain information integration for pan-sharpening,” in European conference on computer vision . Springer, 2022, pp. 274–

  18. [26]

    Image restoration via frequency selection,

    Y . Cui, W. Ren, X. Cao, and A. Knoll, “Image restoration via frequency selection,” IEEE Transactions on Pattern Analysis and Machine Intelli- gence, 2023. 3

  19. [27]

    Squeeze-and-excitation networks,

    J. Hu, L. Shen, and G. Sun, “Squeeze-and-excitation networks,” in Proceedings of the IEEE conference on computer vision and pattern recognition, 2018, pp. 7132–7141. 3, 7, 13

  20. [28]

    Fcanet: Frequency channel attention networks,

    Z. Qin, P. Zhang, F. Wu, and X. Li, “Fcanet: Frequency channel attention networks,” in Proceedings of the IEEE/CVF international conference on computer vision, 2021, pp. 783–792. 3

  21. [29]

    Spanet: Frequency- balancing token mixer using spectral pooling aggregation modulation,

    G. Yun, J. Yoo, K. Kim, J. Lee, and D. H. Kim, “Spanet: Frequency- balancing token mixer using spectral pooling aggregation modulation,” in Proceedings of the IEEE/CVF International Conference on Computer Vision, 2023, pp. 6113–6124. 3

  22. [30]

    Detecting cam- ouflaged object in frequency domain,

    Y . Zhong, B. Li, L. Tang, S. Kuang, S. Wu, and S. Ding, “Detecting cam- ouflaged object in frequency domain,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2022, pp. 4504–

  23. [31]

    Britanak, P

    V . Britanak, P. C. Yip, and K. R. Rao, Discrete cosine and sine trans- forms: general properties, fast algorithms and integer approximations . Elsevier, 2010. 3, 7

  24. [32]

    Frequency-spatial entanglement learning for camouflaged object detection,

    Y . Sun, C. Xu, J. Yang, H. Xuan, and L. Luo, “Frequency-spatial entanglement learning for camouflaged object detection,” in European Conference on Computer Vision . Springer, 2024, pp. 343–360. 3, 4

  25. [33]

    Gather-excite: Exploiting feature context in convolutional neural networks,

    J. Hu, L. Shen, S. Albanie, G. Sun, and A. Vedaldi, “Gather-excite: Exploiting feature context in convolutional neural networks,” Advances in neural information processing systems , vol. 31, 2018. 3

  26. [34]

    Cbam: Convolutional block attention module,

    S. Woo, J. Park, J.-Y . Lee, and I. S. Kweon, “Cbam: Convolutional block attention module,” in Proceedings of the European conference on computer vision (ECCV) , 2018, pp. 3–19. 3

  27. [35]

    Coordinate attention for efficient mobile network design,

    Q. Hou, D. Zhou, and J. Feng, “Coordinate attention for efficient mobile network design,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , 2021, pp. 13 713–13 722. 3

  28. [36]

    Bam: Bottleneck attention module,

    J. Park, S. Woo, J.-Y . Lee, and I. S. Kweon, “Bam: Bottleneck attention module,” arXiv preprint arXiv:1807.06514 , 2018. 3

  29. [37]

    Dual attention network for scene segmentation,

    J. Fu, J. Liu, H. Tian, Y . Li, Y . Bao, Z. Fang, and H. Lu, “Dual attention network for scene segmentation,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2019, pp. 3146–

  30. [38]

    Global second-order pooling convolutional networks,

    Z. Gao, J. Xie, Q. Wang, and P. Li, “Global second-order pooling convolutional networks,” in Proceedings of the IEEE/CVF Conference on computer vision and pattern recognition , 2019, pp. 3024–3033. 3

  31. [39]

    Selective kernel networks,

    X. Li, W. Wang, X. Hu, and J. Yang, “Selective kernel networks,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2019, pp. 510–519. 3

  32. [40]

    Resnest: Split-attention networks,

    H. Zhang, C. Wu, Z. Zhang, Y . Zhu, H. Lin, Z. Zhang, Y . Sun, T. He, J. Mueller, R. Manmatha et al. , “Resnest: Split-attention networks,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2022, pp. 2736–2746. 3

  33. [41]

    Attention is all you need,

    A. Vaswani, N. Shazeer, N. Parmar, J. Uszkoreit, L. Jones, A. N. Gomez, Ł. Kaiser, and I. Polosukhin, “Attention is all you need,” Advances in neural information processing systems , vol. 30, 2017. 3, 6, 12

  34. [42]

    Tokens-to-token vit: Training vision transformers from scratch on imagenet,

    L. Yuan, Y . Chen, T. Wang, W. Yu, Y . Shi, Z.-H. Jiang, F. E. Tay, J. Feng, and S. Yan, “Tokens-to-token vit: Training vision transformers from scratch on imagenet,” in Proceedings of the IEEE/CVF international conference on computer vision , 2021, pp. 558–567. 3, 4

  35. [43]

    Pyramid vision transformer: A versatile backbone for dense prediction without convolutions,

    W. Wang, E. Xie, X. Li, D.-P. Fan, K. Song, D. Liang, T. Lu, P. Luo, and L. Shao, “Pyramid vision transformer: A versatile backbone for dense prediction without convolutions,” in Proceedings of the IEEE/CVF international conference on computer vision , 2021, pp. 568–578. 3

  36. [44]

    End-to-end object detection with transformers,

    N. Carion, F. Massa, G. Synnaeve, N. Usunier, A. Kirillov, and S. Zagoruyko, “End-to-end object detection with transformers,” in European conference on computer vision . Springer, 2020, pp. 213–

  37. [45]

    Deformable {detr}: Deformable transformers for end-to-end object detection,

    X. Zhu, W. Su, L. Lu, B. Li, X. Wang, and J. Dai, “Deformable {detr}: Deformable transformers for end-to-end object detection,” in International Conference on Learning Representations , 2021. [Online]. Available: https://openreview.net/forum?id=gZ9hCDWe6ke 4, 9, 14

  38. [46]

    Conditional convolutions for instance segmentation,

    Z. Tian, C. Shen, and H. Chen, “Conditional convolutions for instance segmentation,” in Computer Vision–ECCV 2020: 16th European Con- ference, Glasgow, UK, August 23–28, 2020, Proceedings, Part I 16 . Springer, 2020, pp. 282–298. 4, 9, 10, 11

  39. [47]

    Cascade r-cnn: High quality object detection and instance segmentation,

    Z. Cai and N. Vasconcelos, “Cascade r-cnn: High quality object detection and instance segmentation,” IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 43, no. 5, pp. 1483–1498, 2021. 4, 9, 10, 11

  40. [48]

    Per-pixel classification is not all you need for semantic segmentation,

    B. Cheng, A. Schwing, and A. Kirillov, “Per-pixel classification is not all you need for semantic segmentation,” Advances in neural information processing systems, vol. 34, pp. 17 864–17 875, 2021. 4

  41. [49]

    Masked-attention mask transformer for universal image segmentation,

    B. Cheng, I. Misra, A. G. Schwing, A. Kirillov, and R. Girdhar, “Masked-attention mask transformer for universal image segmentation,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2022, pp. 1290–1299. 4, 9, 10, 11

  42. [50]

    Swin transformer: Hierarchical vision transformer using shifted windows,

    Z. Liu, Y . Lin, Y . Cao, H. Hu, Y . Wei, Z. Zhang, S. Lin, and B. Guo, “Swin transformer: Hierarchical vision transformer using shifted windows,” in Proceedings of the IEEE/CVF international conference on computer vision, 2021, pp. 10 012–10 022. 4, 6, 8, 9

  43. [51]

    Efficientvit: Memory efficient vision transformer with cascaded group attention,

    X. Liu, H. Peng, N. Zheng, Y . Yang, H. Hu, and Y . Yuan, “Efficientvit: Memory efficient vision transformer with cascaded group attention,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2023, pp. 14 420–14 430. 4

  44. [52]

    Pvt v2: Improved baselines with pyramid vision transformer,

    W. Wang, E. Xie, X. Li, D.-P. Fan, K. Song, D. Liang, T. Lu, P. Luo, and L. Shao, “Pvt v2: Improved baselines with pyramid vision transformer,” Computational visual media , vol. 8, no. 3, pp. 415–424, 2022. 4

  45. [53]

    Towards real-world x-ray security inspection: A high-quality benchmark and lateral inhibition module for prohibited items detection,

    R. Tao, Y . Wei, X. Jiang, H. Li, H. Qin, J. Wang, Y . Ma, L. Zhang, and X. Liu, “Towards real-world x-ray security inspection: A high-quality benchmark and lateral inhibition module for prohibited items detection,” in Proceedings of the IEEE/CVF International Conference on Co...

  46. [54]

    Pidray: A large-scale x-ray benchmark for real-world prohibited item detection,

    L. Zhang, L. Jiang, R. Ji, and H. Fan, “Pidray: A large-scale x-ray benchmark for real-world prohibited item detection,” arXiv preprint arXiv:2211.10763, 2022. 4

  47. [55]

    Towards dual- view x-ray baggage inspection: A large-scale benchmark and adaptive hierarchical cross refinement for prohibited item discovery,

    B. Ma, T. Jia, M. Li, S. Wu, H. Wang, and D. Chen, “Towards dual- view x-ray baggage inspection: A large-scale benchmark and adaptive hierarchical cross refinement for prohibited item discovery,”IEEE Trans- actions on Information Forensics and Security , pp. 1–1, 2024. 4

  48. [56]

    Detection of novel prohibited item categories for real-world security inspection,

    S. Lin, T. Jia, H. Wang, B. Ma, M. Li, and D. Chen, “Detection of novel prohibited item categories for real-world security inspection,” Engineering Applications of Artificial Intelligence , vol. 144, p. 110110,

  49. [57]

    Dino: Detr with improved denoising anchor boxes for end-to- end object detection,

    H. Zhang, F. Li, S. Liu, L. Zhang, H. Su, J. Zhu, L. M. Ni, and H.-Y . Shum, “Dino: Detr with improved denoising anchor boxes for end-to- end object detection,” arXiv preprint arXiv:2203.03605, 2022. 4, 9, 10, 11

  50. [58]

    Mmcl: Boost- ing deformable detr-based detectors with multi-class min-margin con- trastive learning for superior prohibited item detection,

    M. Li, T. Jia, H. Lu, B. Ma, H. Wang, and D. Chen, “Mmcl: Boost- ing deformable detr-based detectors with multi-class min-margin con- trastive learning for superior prohibited item detection,” arXiv preprint arXiv:2406.03176, 2024. 4, 9, 10

  51. [59]

    Cspcl: Category semantic prior contrastive learning for deformable detr-based prohibited item detectors,

    ——, “Cspcl: Category semantic prior contrastive learning for deformable detr-based prohibited item detectors,” arXiv preprint arXiv:2501.16665, 2025. 4, 9, 10

  52. [60]

    R. C. Gonzalez and R. E. Woods, Digital Image Processing (3rd Edition). USA: Prentice-Hall, Inc., 2006. 4, 5

  53. [61]

    A general spatial-frequency learning framework for multimodal image fusion,

    M. Zhou, J. Huang, K. Yan, D. Hong, X. Jia, J. Chanussot, and C. Li, “A general spatial-frequency learning framework for multimodal image fusion,” IEEE Transactions on Pattern Analysis and Machine Intelligence, pp. 1–18, 2024. 4

  54. [62]

    Frequency perception network for camouflaged object detection,

    R. Cong, M. Sun, S. Zhang, X. Zhou, W. Zhang, and Y . Zhao, “Frequency perception network for camouflaged object detection,” in Proceedings of the 31st ACM international conference on multimedia , 2023, pp. 1179–1189. 6

  55. [63]

    Imagenet: A large-scale hierarchical image database,

    J. Deng, W. Dong, R. Socher, L.-J. Li, K. Li, and L. Fei-Fei, “Imagenet: A large-scale hierarchical image database,” in 2009 IEEE conference on computer vision and pattern recognition . Ieee, 2009, pp. 248–255. 8

  56. [64]

    Deep residual learning for image recognition,

    K. He, X. Zhang, S. Ren, and J. Sun, “Deep residual learning for image recognition,” in Proceedings of the IEEE conference on computer vision and pattern recognition , 2016, pp. 770–778. 8, 9

  57. [65]

    Aggregated residual transformations for deep neural networks,

    S. Xie, R. Girshick, P. Doll ´ar, Z. Tu, and K. He, “Aggregated residual transformations for deep neural networks,” in Proceedings of the IEEE conference on computer vision and pattern recognition, 2017, pp. 1492–

  58. [66]

    MMDetection: Open mmlab detection toolbox and benchmark,

    K. Chen, J. Wang, J. Pang, Y . Cao, Y . Xiong, X. Li, S. Sun, W. Feng, Z. Liu, J. Xu, Z. Zhang, D. Cheng, C. Zhu, T. Cheng, Q. Zhao, B. Li, X. Lu, R. Zhu, Y . Wu, J. Dai, J. Wang, J. Shi, W. Ouyang, C. C. Loy, and D. Lin, “MMDetection: Open mmlab detection toolbox and benchmar...

  59. [67]

    Microsoft coco: Common objects in context,

    T.-Y . Lin, M. Maire, S. Belongie, J. Hays, P. Perona, D. Ramanan, P. Doll ´ar, and C. L. Zitnick, “Microsoft coco: Common objects in context,” in Computer Vision–ECCV 2014: 13th European Conference, Zurich, Switzerland, September 6-12, 2014, Proceedings, Part V 13 . Springer,...

  60. [68]

    The pascal visual object classes (voc) challenge,

    M. Everingham, L. Van Gool, C. K. Williams, J. Winn, and A. Zisser- man, “The pascal visual object classes (voc) challenge,” International journal of computer vision , vol. 88, pp. 303–338, 2010. 9

  61. [69]

    Detrs beat yolos on real-time object detection,

    Y . Zhao, W. Lv, S. Xu, J. Wei, G. Wang, Q. Dang, Y . Liu, and J. Chen, “Detrs beat yolos on real-time object detection,” arXiv preprint arXiv:2304.08069, 2023. 9, 10, 11

  62. [70]

    Mask r-cnn,

    K. He, G. Gkioxari, P. Doll ´ar, and R. Girshick, “Mask r-cnn,” in Proceedings of the IEEE international conference on computer vision , 2017, pp. 2961–2969. 9, 10, 11

  63. [71]

    Faster r-cnn: Towards real-time object detection with region proposal networks,

    S. Ren, K. He, R. Girshick, and J. Sun, “Faster r-cnn: Towards real-time object detection with region proposal networks,” IEEE Transactions on Pattern Analysis & Machine Intelligence, vol. 39, no. 06, pp. 1137–1149,

  64. [72]

    Cascade r-cnn: Delving into high quality object detection,

    Z. Cai and N. Vasconcelos, “Cascade r-cnn: Delving into high quality object detection,” in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR) , June 2018. 10, 11

  65. [73]

    Bridging the gap between anchor-based and anchor-free detection via adaptive training sample selection,

    S. Zhang, C. Chi, Y . Yao, Z. Lei, and S. Z. Li, “Bridging the gap between anchor-based and anchor-free detection via adaptive training sample selection,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , 2020, pp. 9759–9768. 10, 11

  66. [74]

    Generalized focal loss: Learning qualified and distributed bounding boxes for dense object detection,

    X. Li, W. Wang, L. Wu, S. Chen, X. Hu, J. Li, J. Tang, and J. Yang, “Generalized focal loss: Learning qualified and distributed bounding boxes for dense object detection,” Advances in Neural Information Processing Systems, vol. 33, pp. 21 002–21 012, 2020. 10, 11

  67. [75]

    Solo: Segmenting objects by locations,

    X. Wang, T. Kong, C. Shen, Y . Jiang, and L. Li, “Solo: Segmenting objects by locations,” in Computer Vision–ECCV 2020: 16th European Conference, Glasgow, UK, August 23–28, 2020, Proceedings, Part XVIII

  68. [76]

    Springer, 2020, pp. 649–665. 10, 11

  69. [77]

    Solov2: Dynamic and fast instance segmentation,

    X. Wang, R. Zhang, T. Kong, L. Li, and C. Shen, “Solov2: Dynamic and fast instance segmentation,” Advances in Neural information processing systems, vol. 33, pp. 17 721–17 732, 2020. 10, 11

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.