Pith. sign in

REVIEW 4 major objections 4 minor 38 references

The paper argues that DETR's self-attention dilutes infrared small-target embeddings and shows that a Fourier-spectrum query initialization recovers them, setting a new state of the art on three benchmarks.

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 ↗

Frequency-guided query initialization (Fourier patch screening, dynamic embedding enhancement, reliability-consistency fusion) achieves SOTA infrared small-target detection on three public benchmarks with negligible added compute.

T0 review reviewed 2026-08-04 challenge →

load-bearing objection A plausible and cheap frequency-guided query initialization for IRSTD, but the submission has internal inconsistencies and a key premise needs quantitative support. the 4 major comments →

arxiv 2601.02837 v2 pith:C4ARI3MX submitted 2026-01-06 cs.CV cs.AI

Breaking Self-Attention Failure: Rethinking Query Initialization for Infrared Small Target Detection

Yuteng Liu , Duanni Meng , Yimian Dai , Maoxun Yuan , Xingxing Wei , Bo li This is my paper
classification cs.CV cs.AI
keywords infrared small target detectionDETR query initializationself-attention embedding dilutionFourier spectrum screeningfrequency-guided detectiontiny object detectionfeature enhancementquery fusion
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.

The reading

Self-attention in DETR-style detectors mixes the few tokens that represent a tiny infrared target with the overwhelming number of background tokens, so the target's identity gets diluted before queries are initialized. SEF-DETR fixes this by screening local patches with their full Fourier magnitude spectrum, building a target-relevant density map, amplifying encoder features in those regions, and fusing spatial and frequency confidence to select queries. On IRSTD-1k, it beats the DINO baseline by 1.8 AP and improves very-tiny-target APvt from 29.6 to 32.8, with only 0.27M extra parameters and 0.08 GFLOPs. If correct, this gives DETR-based infrared small-target detection a principled, cheap query-initialization remedy rather than a full architectural redesign.

Core claim

The central claim is that the failure of DETR-based detectors on infrared small targets is not an architecture problem but a query-initialization problem: because attention weights are normalized and background tokens vastly outnumber target tokens, every aggregated embedding becomes dominated by background features. SEF-DETR corrects this by using the complete Fourier spectrum of overlapping local patches to produce a target-relevant density map, then uses that map to enhance encoder features in target-like regions (DEE) and to re-rank candidate queries by spatial-frequency consistency and reliability (RCF). Experiments on IRSTD-1k, NUAA-SIRST, and NUDT-SIRST report the best precision/recal

What carries the argument

The load-bearing object is the patch-wise frequency density map S_freq produced by the Frequency-guided Patch Screening module: the magnitude spectrum of each overlapping local patch is flattened and passed through an MLP with a classification head, and overlapping predictions are aggregated by geometric mean. This map drives both Dynamic Embedding Enhancement, which modulates encoder features as Q' = Q ⊙ (1 + M) using a learnable threshold, and Reliability-Consistency-aware Fusion, which computes S_final = S_spatial · (1 + C · (1 + R)) with consistency C = 1 − |S_spatial − S_freq| and reliability R = 2·|S_freq − 0.5|. All gains flow through this map, making it the central mechanism carrying

Load-bearing premise

The FPS module's premise that the full Fourier magnitude spectrum of local patches reliably separates true small targets from background clutter and target-like distractors is the load-bearing claim; if those spectra overlap in realistic infrared scenes, the density map misranks queries and the subsequent enhancement and fusion inherit the error.

What would settle it

Measure the area under the ROC curve of the FPS classifier on a held-out set of labeled true-target patches versus labeled distractor and background patches from IRSTD-1k; an AUC at or near chance would show the frequency screen cannot carry the reported detection gains.

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

If this is right

  • DETR-based infrared small-target detectors should initialize queries from frequency-domain priors rather than relying on encoder token responses alone.
  • Very tiny targets (under 82 pixels) benefit most, with APvt rising from 29.6 to 32.8 on IRSTD-1k, suggesting the fix addresses the most challenging scale.
  • The added cost is minuscule (+0.27M parameters, +0.08 GFLOPs), so the frequency-screening components can be retrofitted into existing DETR pipelines without heavy re-engineering.
  • Using the complete Fourier spectrum outperforms using only high- or low-frequency bands, indicating that complementary discriminative information exists across bands.
  • The fusion of reliability and consistency terms is what unlocks the gain; naive additive fusion of spatial and frequency scores yields far smaller improvements.

Where Pith is reading between the lines

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

  • The frequency-screening principle may transfer to other tiny-object or weak-signal detection domains (e.g., aerial or satellite imagery) where background tokens similarly dominate self-attention.
  • A quantitative separability analysis—such as the AUC of the FPS classifier on held-out true-target versus distractor patches—would turn the paper's central premise into a directly testable claim.
  • The claim that dilution is 'inevitable' rests on softmax normalization and equal token weighting; alternative attention designs that prune tokens or mask background might avoid the problem without frequency priors.
  • The manually designed fusion formula and learnable threshold could likely be replaced by a single learned module, which would simplify the pipeline if the gains persist.
Share X Bluesky LinkedIn Reddit HN

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

4 major / 4 minor

Summary. This paper addresses the poor performance of DETR-family detectors on infrared small target detection. It argues in Sec. 3.1 (Eqs. 1–7) that target embeddings are diluted by dominant background tokens through softmax self-attention, and proposes SEF-DETR, a DINO-based framework with three modules: FPS, which classifies overlapping local patches from their Fourier magnitude spectra and aggregates them into a target-relevant density map; DEE, which masks and amplifies encoder features where the density map exceeds a threshold; and RCF, which fuses spatial confidence with frequency reliability and consistency to select the top-K queries. Experiments compare against CNN- and DETR-based baselines on NUAA-SIRST, NUDT-SIRST, and IRSTD-1k, reporting best F1/P/R in Table 1 and best AI-TOD AP metrics in Table 2 at +0.27M parameters and +0.08 GFLOPs over DINO.

Significance. If verified, SEF-DETR is a practically useful, low-cost modification that turns a generic DETR detector into a strong IRSTD detector, and the frequency-guided query screening idea could generalize to other small-object tasks. The evaluation has solid scaffolding: three public datasets with held-out splits, comparisons to competitive baselines, and component-wise ablations (Tables 3–6). The FPS classifier is trained with ground-truth patch occupancy, not fitted to the final metric, which reduces circularity. The main risk is that the central premise—Fourier magnitude spectra separate targets from target-like distractors—is not directly quantified; and several reporting inconsistencies must be corrected.

major comments (4)
  1. [Sec. 4.3 / Table 2] The prose reports '38.3% AP, 85.0% AP50, 27.0% AP75, 31.3% APvt, 49.9% APt, 59.6% APs, and 61.7% APm' for SEF-DETR, but Table 2's SEF-DETR row reports 38.9, 86.7, 27.1, 32.8, 50.8, 56.1 and contains no APm column. Only the Table 2 numbers are consistent with the abstract (AP 38.9 vs. DINO 37.1). The discrepancy must be resolved; as it stands, the claimed numerical improvements in the text are not reproducible from the tables.
  2. [Sec. 3.2, Eqs. (8)–(10)] The whole mechanism passes through the FPS density map S_freq, yet the claim that Fourier magnitude spectra distinguish true targets from target-like distractors is not quantitatively established. Fig. 2 is a single illustrative example; Table 4 shows only final AP for full vs. band-limited spectra, which does not isolate FPS classifier accuracy. Please provide patch-level precision/recall, ROC/AUC, or distributions of S_freq at true target centers vs. distractor locations, and an ablation replacing the magnitude spectrum with a non-frequency feature of equal complexity. Without this, the reported +1.8 AP could derive from added parameters or feature modulation rather than from spectral discriminability.
  3. [Sec. 3.2, Eq. (11)] The DEE mask uses a 'learnable threshold a' followed by a hard binarization; if a is trained by gradient descent, the paper must state the gradient rule (straight-through, stop-gradient, etc.) and update details. As written, the mechanism is not reproducible. This is load-bearing because DEE is one of the two consumers of S_freq, and Table 3 attributes part of the gain to it.
  4. [Secs. 3.2 and 4.2] Several parameters that determine the FPS operating point are unstated: patch size p, sliding-window stride, number of overlapping patches n, and top-K query count. The FPS loss weight lambda is given, but not the initial value or schedule of the learnable threshold. Please report these values and, ideally, sensitivity of the final AP to p/stride/K; otherwise the method cannot be replicated.
minor comments (4)
  1. [Table 1, MDvsFA rows] The reported F1 values (47.5, 59.7, 26.2) do not match 2PR/(P+R) computed from the listed P/R values (51.4, 63.4, 29.2). Clarify how F1 is computed or correct the entries.
  2. [Sec. 3.2, Eq. (8)] The text calls the input the 'complete Fourier spectrum', but only the magnitude |F_j| is used; the phase is not used. Clarify this in the description.
  3. [Sec. 4.3] The APm value is mentioned in the prose but never defined in Sec. 4.1 and absent from Table 2. Either add the corresponding column/definition or remove the number.
  4. [Sec. 3.1, Fig. 3] Please specify how M, Ω_t, and Ω_b are computed for the similarity measurements, and show error bars across images/files. The current curves lack confidence intervals.

Circularity Check

0 steps flagged

No circularity; the derivation chain is supervised, benchmarked against held-out public data, and the frequency prior is not fitted to the final metric.

full rationale

The paper's central mechanism is not circular. FPS is a supervised classifier (Eqs. 8-9, trained with Eq. 16) whose labels are ground-truth patch occupancy; its density map S_freq is an intermediate prior, not a direct copy of the test labels. DEE and RCF use this map to modulate features and re-rank queries, and final detection quality is measured end-to-end against held-out splits of three public datasets (3:1:1 split, Sec. 4.1) with external DETR/CNN baselines. No parameter is fitted to the reported AP/F1 numbers, and no load-bearing claim rests on a self-citation: ref [31] is a comparison baseline, not evidence for the method. The Eq. 4 'embedding dilution' argument and the FPS spectral-separability premise are empirical-support concerns (the softmax conclusion is not forced by the equation, and no separability analysis is given), but they are not cases where a result reduces by construction or by definition to its inputs. Hence score 0.

Axiom & Free-Parameter Ledger

5 free parameters · 4 axioms · 0 invented entities

The paper's contribution is a three-module empirical system. The honest ledger: five design choices/numbers are set by hand or left unspecified (patch geometry, learnable mask threshold with an unclear gradient path, fusion constants, loss weight, top-K), and four domain assumptions are needed for the analysis and comparisons. No invented entities. This is typical for an applied deep-learning paper; nothing is pulled from a hat, but the hand-set values plus the unstated comparison protocol bound how much a reader can verify.

free parameters (5)
  • FPS patch size p and sliding-window stride = not reported
    Chosen by hand (Sec 3.2, Eq. 8); they set the resolution of the target-relevant density map and are neither documented nor ablated.
  • Learnable threshold a in DEE mask (Eq. 11) = learned, value not reported
    A learnable scalar threshold over a hard 0/1 mask; the gradient path is unspecified (no straight-through estimator or soft relaxation described), so it is unclear how a is fitted.
  • RCF fusion constants (0.5, 1, 2) = 0.5, 1, 2 (Eqs. 13-15)
    Hand-designed coefficients for consistency/reliability amplification; only coarse fusion strategies are ablated (Table 6), not the constants themselves.
  • Loss weight lambda = 2 (Eq. 17)
    Hand-chosen balance between Hungarian loss and the patch-wise frequency loss; no ablation of lambda is provided.
  • Top-K query count for decoder = not reported
    Number of candidate queries passed to the decoder after RCF ranking is not stated, though it directly affects the precision/recall trade-off.
axioms (4)
  • domain assumption The Fourier magnitude spectrum of a local patch separates true IRSTs from target-like false alarms and background (Sec 1, Fig 2; Eqs. 8-9).
    The entire FPS module rests on this; it is supported only by a qualitative illustration, with no quantitative separability analysis.
  • ad hoc to paper Attention weights in DETR are diffuse enough that background tokens dominate the output embeddings of target queries (needed for Eq. 4's conclusion).
    Eq. 4 derives background dominance from |Omega_t| << |Omega_b| plus softmax normalization, but dominance also requires attention not be sharply peaked on targets; this premise is what the analysis is meant to establish.
  • domain assumption Random 3:1:1 splits with balanced target distribution match the splits used by the prior published baselines in Tables 1-2.
    The paper does not state whether baselines were retrained under the same split; cross-paper numbers on different splits are not strictly comparable.
  • domain assumption The P/R/F1 protocol for Table 1 applies a consistent (unstated) confidence threshold to SEF-DETR's query outputs.
    Precision/recall requires thresholding DETR outputs; the threshold is not specified, so 100.0% precision on NUDT-SIRST cannot be reproduced or audited.

reviewed 2026-08-04 · how reviews work

0 comments
Cite this review

Pith. "Pith review of Breaking Self-Attention Failure: Rethinking Query Initialization for Infrared Small Target Detection." pith.science (2026). https://pith.science/paper/C4ARI3MX

@misc{pith2026260102837,
  author       = {Pith},
  title        = {Pith review of: Breaking Self-Attention Failure: Rethinking Query Initialization for Infrared Small Target Detection},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/C4ARI3MX}},
  note         = {Machine review of arXiv:2601.02837}
}
Share X Bluesky LinkedIn Reddit HN
read the original abstract

Infrared small target detection (IRSTD) faces significant challenges due to low signal-to-noise ratios, extremely small target sizes, and complex cluttered backgrounds. Although DETR-based detectors benefit from global context modeling, their query initialization can become unreliable in IRSTD because only a few encoder tokens correspond to targets, while the majority describe the background. We revisit this phenomenon and reveal that the target-relevant embeddings of IRST are inevitably overwhelmed by dominant background features due to the self-attention mechanism, leading to unreliable query initialization and inaccurate target localization. To address this issue, we propose SEF-DETR, a novel framework that refines query initialization through Patch-wise Spectral Screening (PSS), Frequency-Routed Examination (FRE), and Reliability-Consistency-aware Fusion (RCF). PSS encodes the radial and directional energy distributions of local Fourier spectra and aggregates overlapping patch predictions into a target-relevant density map. Guided by this map, FRE performs sparse deformable re-examination after each encoder layer, while retaining content-driven sampling to distinguish true targets from frequency-domain false alarms. RCF further re-ranks candidate queries according to spatial-frequency consistency and frequency reliability. Extensive experiments on three public IRSTD datasets demonstrate that SEF-DETR achieves superior detection performance over state-of-the-art methods with low computational overhead, providing a robust and efficient DETR-based solution for IRSTD.

Figures

Figures reproduced from arXiv: 2601.02837 by Bo Li, Duanni Meng, Maoxun Yuan, Xingxing Wei, Yimian Dai, Yuteng Liu.

Figure 1
Figure 1. Figure 1: Comparison of three different query initialization meth [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Illustration of a local patch of IRST from IRSTD-1k [PITH_FULL_IMAGE:figures/full_fig_p002_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Comparison of similarity s in the “l = 0 to l = 6” layers on IRSTD-1k, NUAA-SIRST and NUDT-SIRST dataset. It can be seen that the target-relevant embedding in the deeper layers are gradually diluted by the background-relevant embedding. 3. Method 3.1. Analysis ❶ Revisit self-attention in DETR. Given an input infrared image feature map X ∈ R H×W×C , where H, W, and C denote the spatial resolution and channe… view at source ↗
Figure 4
Figure 4. Figure 4: Overview of our proposed SEF-DETR. The input infrared image is processed through two complementary paths. The top branch [PITH_FULL_IMAGE:figures/full_fig_p004_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Visualization comparison of detection results via different methods on representative images from IRSTD-1k datasets, indicate the [PITH_FULL_IMAGE:figures/full_fig_p007_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Visualization of the spatial confidence maps from the [PITH_FULL_IMAGE:figures/full_fig_p007_6.png] view at source ↗

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

38 extracted references · 7 linked inside Pith

  1. [1]

    Layer normalization.arXiv preprint arXiv:1607.06450,

    Jimmy Lei Ba, Jamie Ryan Kiros, and Geoffrey E Hin- ton. Layer normalization.arXiv preprint arXiv:1607.06450,

    Pith/arXiv arXiv

  2. [2]

    End-to- end object detection with transformers

    Nicolas Carion, Francisco Massa, Gabriel Synnaeve, Nicolas Usunier, Alexander Kirillov, and Sergey Zagoruyko. End-to- end object detection with transformers. InEuropean confer- ence on computer vision, pages 213–229. Springer, 2020. 2, 5

    2020

  3. [3]

    Tci-former: Thermal conduction-inspired transformer for infrared small target detection

    Tianxiang Chen, Zhentao Tan, Qi Chu, Yue Wu, Bin Liu, and Nenghai Yu. Tci-former: Thermal conduction-inspired transformer for infrared small target detection. InProceed- ings of the AAAI Conference on Artificial Intelligence, pages 1201–1209, 2024. 3

    2024

  4. [4]

    Asymmetric contextual modulation for infrared small tar- get detection

    Yimian Dai, Yiquan Wu, Fei Zhou, and Kobus Barnard. Asymmetric contextual modulation for infrared small tar- get detection. InProceedings of the IEEE/CVF winter con- ference on applications of computer vision, pages 950–959,

  5. [5]

    At- tentional local contrast networks for infrared small target de- tection.IEEE transactions on geoscience and remote sens- ing, 59(11):9813–9824, 2021

    Yimian Dai, Yiquan Wu, Fei Zhou, and Kobus Barnard. At- tentional local contrast networks for infrared small target de- tection.IEEE transactions on geoscience and remote sens- ing, 59(11):9813–9824, 2021. 1, 6

    2021

  6. [6]

    One-stage cascade refinement networks for infrared small target detection.IEEE transactions on geo- science and remote sensing, 61:1–17, 2023

    Yimian Dai, Xiang Li, Fei Zhou, Yulei Qian, Yaohong Chen, and Jian Yang. One-stage cascade refinement networks for infrared small target detection.IEEE transactions on geo- science and remote sensing, 61:1–17, 2023. 1

    2023

  7. [7]

    Deep sparse rectifier neural networks

    Xavier Glorot, Antoine Bordes, and Yoshua Bengio. Deep sparse rectifier neural networks. InProceedings of the four- teenth international conference on artificial intelligence and statistics, pages 315–323. JMLR Workshop and Conference Proceedings, 2011. 5

    2011

  8. [8]

    Deep wavelet prediction for image super- resolution

    Tiantong Guo, Hojjat Seyed Mousavi, Tiep Huu Vu, and Vishal Monga. Deep wavelet prediction for image super- resolution. InProceedings of the IEEE conference on com- puter vision and pattern recognition workshops, pages 104– 113, 2017. 3

    2017

  9. [9]

    Ultralytics yolov8, 2023

    Glenn Jocher, Ayush Chaurasia, and Jing Qiu. Ultralytics yolov8, 2023. 6

    2023

  10. [10]

    Dense nested attention network for infrared small target detection.IEEE Transactions on Image Processing, 32:1745–1758, 2022

    Boyang Li, Chao Xiao, Longguang Wang, Yingqian Wang, Zaiping Lin, Miao Li, Wei An, and Yulan Guo. Dense nested attention network for infrared small target detection.IEEE Transactions on Image Processing, 32:1745–1758, 2022. 2, 6

    2022

  11. [11]

    Dn-detr: Accelerate detr training by intro- ducing query denoising

    Feng Li, Hao Zhang, Shilong Liu, Jian Guo, Lionel M Ni, and Lei Zhang. Dn-detr: Accelerate detr training by intro- ducing query denoising. InProceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 13619–13627, 2022. 2, 3, 6

    2022

  12. [12]

    Hstnet: A hybrid spatial-channel sparse trans- former network for infrared small target detection.IEEE Transactions on Geoscience and Remote Sensing, 2025

    Ke Li, Yining Wang, Fujun Han, Hu Wang, Zige Xiong, and Yan Tian. Hstnet: A hybrid spatial-channel sparse trans- former network for infrared small target detection.IEEE Transactions on Geoscience and Remote Sensing, 2025. 3

    2025

  13. [13]

    Ir-transdet: Infrared dim and small target detection with ir-transformer.IEEE Transactions on Geo- science and Remote Sensing, 61:1–13, 2023

    Jian Lin, Shaoyi Li, Liang Zhang, Xi Yang, Binbin Yan, and Zhongjie Meng. Ir-transdet: Infrared dim and small target detection with ir-transformer.IEEE Transactions on Geo- science and Remote Sensing, 61:1–13, 2023. 3

    2023

  14. [14]

    Focal loss for dense object detection

    Tsung-Yi Lin, Priya Goyal, Ross Girshick, Kaiming He, and Piotr Doll´ar. Focal loss for dense object detection. InPro- ceedings of the IEEE international conference on computer vision, pages 2980–2988, 2017. 5

    2017

  15. [15]

    Infrared small target detection with scale and loca- tion sensitivity

    Qiankun Liu, Rui Liu, Bolun Zheng, Hongkui Wang, and Ying Fu. Infrared small target detection with scale and loca- tion sensitivity. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 17490– 17499, 2024. 2

    2024

  16. [16]

    Dab-detr: Dynamic anchor boxes are better queries for detr.arXiv preprint arXiv:2201.12329, 2022

    Shilong Liu, Feng Li, Hao Zhang, Xiao Yang, Xianbiao Qi, Hang Su, Jun Zhu, and Lei Zhang. Dab-detr: Dynamic anchor boxes are better queries for detr.arXiv preprint arXiv:2201.12329, 2022. 2, 3, 6

    Pith/arXiv arXiv 2022

  17. [17]

    Decoupled weight decay regularization.arXiv preprint arXiv:1711.05101, 2017

    Ilya Loshchilov and Frank Hutter. Decoupled weight decay regularization.arXiv preprint arXiv:1711.05101, 2017. 6

    Pith/arXiv arXiv 2017

  18. [18]

    Scaling the scattering transform: Deep hybrid networks

    Edouard Oyallon, Eugene Belilovsky, and Sergey Zagoruyko. Scaling the scattering transform: Deep hybrid networks. InProceedings of the IEEE international conference on computer vision, pages 5618–5627, 2017. 3

    2017

  19. [19]

    Global filter networks for image classification

    Yongming Rao, Wenliang Zhao, Zheng Zhu, Jiwen Lu, and Jie Zhou. Global filter networks for image classification. Advances in neural information processing systems, 34:980– 993, 2021. 3

    2021

  20. [20]

    Miss detection vs

    Huan Wang, Luping Zhou, and Lei Wang. Miss detection vs. false alarm: Adversarial learning for small object segmen- tation in infrared images. InProceedings of the IEEE/CVF international conference on computer vision, pages 8509– 8518, 2019. 6 9

    2019

  21. [21]

    Tiny object detection in aerial images

    Jinwang Wang, Wen Yang, Haowen Guo, Ruixiang Zhang, and Gui-Song Xia. Tiny object detection in aerial images. In 2020 25th international conference on pattern recognition (ICPR), pages 3791–3798. IEEE, 2021. 6

    2020

  22. [22]

    Eflnet: Enhancing feature learning network for infrared small target detection.IEEE Transac- tions on Geoscience and Remote Sensing, 62:1–11, 2024

    Bo Yang, Xinyu Zhang, Jian Zhang, Jun Luo, Mingliang Zhou, and Yangjun Pi. Eflnet: Enhancing feature learning network for infrared small target detection.IEEE Transac- tions on Geoscience and Remote Sensing, 62:1–11, 2024. 6

    2024

  23. [23]

    Istd-detr: A deep learning algorithm based on detr and super- resolution for infrared small target detection.Neurocomput- ing, 621:129289, 2025

    Huanyu Yang, Jun Wang, Yuming Bo, and Jiacun Wang. Istd-detr: A deep learning algorithm based on detr and super- resolution for infrared small target detection.Neurocomput- ing, 621:129289, 2025. 3

    2025

  24. [24]

    Pinwheel-shaped convolution and scale-based dynamic loss for infrared small target detection

    Jiangnan Yang, Shuangli Liu, Jingjun Wu, Xinyu Su, Nan Hai, and Xueli Huang. Pinwheel-shaped convolution and scale-based dynamic loss for infrared small target detection. InProceedings of the AAAI Conference on Artificial Intelli- gence, pages 9202–9210, 2025. 6

    2025

  25. [25]

    Fda: Fourier domain adaptation for semantic segmentation

    Yanchao Yang and Stefano Soatto. Fda: Fourier domain adaptation for semantic segmentation. InProceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 4085–4095, 2020. 3

    2020

  26. [26]

    Wave-vit: Unifying wavelet and transformers for visual representation learning

    Ting Yao, Yingwei Pan, Yehao Li, Chong-Wah Ngo, and Tao Mei. Wave-vit: Unifying wavelet and transformers for visual representation learning. InEuropean conference on computer vision, pages 328–345. Springer, 2022. 3

    2022

  27. [27]

    C 2former: Calibrated and complementary transformer for rgb-infrared object de- tection.IEEE Transactions on Geoscience and Remote Sens- ing, 62:1–12, 2024

    Maoxun Yuan and Xingxing Wei. C 2former: Calibrated and complementary transformer for rgb-infrared object de- tection.IEEE Transactions on Geoscience and Remote Sens- ing, 62:1–12, 2024. 1

    2024

  28. [28]

    Transla- tion, scale and rotation: Cross-modal alignment meets rgb- infrared vehicle detection

    Maoxun Yuan, Yinyan Wang, and Xingxing Wei. Transla- tion, scale and rotation: Cross-modal alignment meets rgb- infrared vehicle detection. InEuropean Conference on Com- puter Vision, pages 509–525. Springer, 2022. 1

    2022

  29. [29]

    Improving rgb-infrared object detection with cascade alignment-guided transformer.Information Fusion, 105:102246, 2024

    Maoxun Yuan, Xiaorong Shi, Nan Wang, Yinyan Wang, and Xingxing Wei. Improving rgb-infrared object detection with cascade alignment-guided transformer.Information Fusion, 105:102246, 2024. 1

    2024

  30. [30]

    Unirgb-ir: A unified frame- work for visible-infrared semantic tasks via adapter tuning

    Maoxun Yuan, Bo Cui, Tianyi Zhao, Jiayi Wang, Shan Fu, Xue Yang, and Xingxing Wei. Unirgb-ir: A unified frame- work for visible-infrared semantic tasks via adapter tuning. InProceedings of the 33rd ACM International Conference on Multimedia, pages 2409–2418, 2025. 1

    2025

  31. [31]

    Ns-fpn: Improving infrared small target detection and segmentation from noise suppression perspective.arXiv preprint arXiv:2508.06878,

    Maoxun Yuan, Duanni Meng, Ziteng Xi, Tianyi Zhao, Shiji Zhao, Yimian Dai, and Xingxing Wei. Ns-fpn: Improving infrared small target detection and segmentation from noise suppression perspective.arXiv preprint arXiv:2508.06878,

    arXiv

  32. [32]

    Sctransnet: Spatial-channel cross transformer net- work for infrared small target detection.IEEE Transactions on Geoscience and Remote Sensing, 62:1–15, 2024

    Shuai Yuan, Hanlin Qin, Xiang Yan, Naveed Akhtar, and Aj- mal Mian. Sctransnet: Spatial-channel cross transformer net- work for infrared small target detection.IEEE Transactions on Geoscience and Remote Sensing, 62:1–15, 2024. 3

    2024

  33. [33]

    Dino: Detr with improved denoising anchor boxes for end-to-end object detection.arXiv preprint arXiv:2203.03605, 2022

    Hao Zhang, Feng Li, Shilong Liu, Lei Zhang, Hang Su, Jun Zhu, Lionel M Ni, and Heung-Yeung Shum. Dino: Detr with improved denoising anchor boxes for end-to-end object detection.arXiv preprint arXiv:2203.03605, 2022. 2, 3, 6

    Pith/arXiv arXiv 2022

  34. [34]

    Isnet: Shape matters for infrared small target detection

    Mingjin Zhang, Rui Zhang, Yuxiang Yang, Haichen Bai, Jing Zhang, and Jie Guo. Isnet: Shape matters for infrared small target detection. InProceedings of the IEEE/CVF con- ference on computer vision and pattern recognition, pages 877–886, 2022. 6

    2022

  35. [35]

    Agpcnet: Attention-guided pyramid context net- works for infrared small target detection.arXiv preprint arXiv:2111.03580, 2021

    Tianfang Zhang, Siying Cao, Tian Pu, and Zhenming Peng. Agpcnet: Attention-guided pyramid context net- works for infrared small target detection.arXiv preprint arXiv:2111.03580, 2021. 6

    Pith/arXiv arXiv 2021

  36. [36]

    Rethinking multi-modal object detection from the perspective of mono-modality fea- ture learning.arXiv preprint arXiv:2503.11780, 2025

    Tianyi Zhao, Boyang Liu, Yanglei Gao, Yiming Sun, Maoxun Yuan, and Xingxing Wei. Rethinking multi-modal object detection from the perspective of mono-modality fea- ture learning.arXiv preprint arXiv:2503.11780, 2025. 1

    Pith/arXiv arXiv 2025

  37. [37]

    Detrs beat yolos on real-time object detection

    Yian Zhao, Wenyu Lv, Shangliang Xu, Jinman Wei, Guanzhong Wang, Qingqing Dang, Yi Liu, and Jie Chen. Detrs beat yolos on real-time object detection. InProceed- ings of the IEEE/CVF conference on computer vision and pattern recognition, pages 16965–16974, 2024. 3

    2024

  38. [38]

    Deformable detr: Deformable trans- formers for end-to-end object detection.arXiv preprint arXiv:2010.04159, 2020

    Xizhou Zhu, Weijie Su, Lewei Lu, Bin Li, Xiaogang Wang, and Jifeng Dai. Deformable detr: Deformable trans- formers for end-to-end object detection.arXiv preprint arXiv:2010.04159, 2020. 2, 6 10

    Pith/arXiv arXiv 2010

This paper was first reviewed by deepseek-v4-flash on August 4, 2026.