Pith. sign in

REVIEW 3 major objections 4 minor 56 references

UniDet-D: A Unified Dynamic Spectral Attention Model for Object Detection under Adverse Weathers

T0 review · 3 major / 4 minor · reviewed 2026-08-07 · deepseek-v4-flash

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

Pith's one-line read UniDet-D claims that a single object detector can handle rain, fog, snow, and low light by learning which DCT frequency bands matter, beating weather-specific multi-task models by 1.2 to 1.4 mAP on a mixed benchmark.

desk verdict A coherent unified adverse-weather detector with a genuinely learnable frequency-gating module, but the unseen-generalization story rests on qualitative evidence and unverified switch behavior. read the letter →

arxiv 2506.12324 v2 pith:UATCJ3WI submitted 2025-06-14 cs.CV

Wei Zhang , Yuantao Wang , Haowei Yang , Yin Zhuang , Shijian Lu , Xuerui Mao This is my paper
classification cs.CV
keywords objectdetectionadverseweatherfrequencylearningdynamicspectralattentiondiscretecosinetransformimagerestorationGumbel-Softmaxdomaingeneralization
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

This paper tries to establish that one end-to-end network can do both image restoration and object detection across diverse adverse weather conditions by learning which frequency bands to keep. The proposed model, UniDet-D, uses a dynamic spectral attention mechanism: a Multi-Spectrum Perception module splits feature channels and projects them onto DCT bases, and an Adaptable Frequency Filtering module learns binary on/off switches for those frequency components. On mixed benchmarks built from PASCAL VOC with synthetic rain, fog, snow, and clean images, UniDet-D reports average mAP gains of 1.18 over TogetherNet and 1.41 over the recent RDMNet baseline. It also reports a 1.85 mAP gain over RDMNet on the unseen real-world ExDark low-light set and qualitatively shows detection under sandstorms and rain-fog mixtures. If these results hold, a single frequency-adaptive detector could replace weather-specific pipelines with small but consistent accuracy gains.

What carries the argument

The load-bearing mechanism is the dynamic spectral perceive-select strategy, made of two modules. MSP (Multi-Spectrum Perception) evenly splits feature channels into n segments and projects each onto a chosen 2D-DCT basis, so the attention vector encodes multiple frequency components instead of only the DC term. AF2 (Adaptable Frequency Filtering) turns those frequency responses into binary switches via a 1x1 convolution producing two logits per component, sampled with Gumbel-Softmax so the selection is differentiable, and regularized by a sparsity-promoting frequency loss. The switches decide which spectral channels are kept, letting the network adapt its frequency emphasis to the degradation present in the image.

What would settle it

Record the binary DCT switches that the trained AF2 module assigns on each synthetic degradation (rain, fog, snow) and compare them to the switches it assigns on unseen real-world sandstorm and rain-fog images. If the retained frequency masks are essentially identical across all conditions, or if they match only the training data's synthesis artifacts, then the claimed generalization to unseen degradations collapses. A quantitative version is to evaluate UniDet-D on real fog benchmarks such as RTTS or real rain benchmarks such as RainCityscapes; if its mAP advantage over RDMNet disappears on those, the spectral gating is not transferring.

Watch

Extended reading notes

Core claim

UniDet-D's core discovery is that a detector can jointly restore and detect by dynamically gating DCT frequency bands per image, rather than relying on global average pooling or fixed frequency subsets. The paper shows that standard channel attention, which reduces each channel to its DC (lowest-frequency) component, discards mid- and high-frequency detail that matters under degradation. Its Multi-Spectrum Perception (MSP) module partitions feature channels and projects each segment onto a designated DCT basis, and the Adaptable Frequency Filtering (AF2) module learns binary on/off switches for those frequency components using Gumbel-Softmax sampling, regularized by a frequency sparsity loss. With this mechanism, the model outperforms per-weather baselines and recent multi-task detectors on every tested degradation type, and it improves on the RDMNet baseline by 1.85 mAP on the unseen real-world ExDark low-light set.

Load-bearing premise

The load-bearing premise is that the on/off frequency choices the model learns from synthetic rain, fog, and snow continue to help on unseen real-world degradations such as sandstorms and rain-fog mixtures, rather than simply memorizing the synthetic artifacts.

Editorial extensions

If this is right

  • A single UniDet-D model trained on a mixed dataset of clean, rain, fog, and snow images can serve all of those conditions without swapping components, according to the Table IV results.
  • Unseen degradation types such as sandstorms and rain-fog mixtures show qualitatively better detection with UniDet-D than with the YOLOXs baseline or RDMNet, according to Figure 5.
  • Training on synthetic low-light data (VLow Light) raises the zero-shot ExDark mAP by 4.22 points, showing the unified model can absorb a new degradation type by fine-tuning.
  • Using the learned frequency switches (AF2) instead of using all DCT bands improves fog-test mAP by 1.54 points, so selective spectral gating is the active ingredient, not merely multi-spectrum projection.

Reading between the lines

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

  • [Editorial inference] If the frequency-switch transfer holds, the same gating mechanism could be applied to other dense prediction tasks such as semantic segmentation or depth estimation under degraded inputs, since those tasks also rely on high-frequency detail.
  • [Editorial inference] A direct way to test the mechanism's claim is to log the activated DCT masks per degradation type during inference; distinct masks for rain, fog, and snow would support the degradation-aware interpretation, while a single fixed mask would indicate the gain comes from elsewhere.
  • [Editorial inference] The generalization evidence is strongest for low light (one real-world set, ExDark) and only qualitative for sandstorms and rain-fog; testing on real fog (RTTS) and real rain (RainCityscapes) would show whether the spectral switches transfer beyond the paper's synthetic distributions.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 4 minor

Summary. The paper proposes UniDet-D, a unified end-to-end network for object detection under adverse weather that jointly performs detection and image restoration. The core technical novelty is a dynamic spectral attention mechanism composed of two modules: MSP, which partitions feature channels and applies per-segment 2D DCT to capture multiple frequency components, and AF2, which learns binary frequency switches via Gumbel-Softmax with a sparsity regularizer. The authors report consistent improvements over YOLOXs, TogetherNet, RDMNet, and several restore-then-detect pipelines on synthetic rain, fog, and snow benchmarks, as well as on the real-world low-light ExDark dataset, and they provide qualitative evidence on unseen sandstorm and rain-fog conditions. The central claim is that a single frequency-adaptive detector can replace weather-specific pipelines with small but consistent mAP gains and generalize to unseen degradation types.

Significance. If the claims hold, UniDet-D would be a useful contribution to degraded-image object detection, combining restoration and detection in one network with an interpretable frequency-selection mechanism. The paper has several strengths: the tables are internally consistent, the ablation in Table VI directly shows that enabling the learned frequency switches improves mAP by 1.54% on VFog, and the evaluation includes external multi-task and restoration-plus-detection baselines rather than only self-comparisons. However, the central generalization claim rests on learned frequency switches whose behavior is never directly analyzed, and the quantitative margins over strong baselines are around 1 mAP with no reported variance, which makes the misstated deltas in the text and the absence of mechanism-level evidence more consequential.

major comments (3)
  1. [IV.C.1 and IV.C.2, Tables I–II] The reported mAP improvements over RDMNet and TogetherNet do not match the tables. Section IV.C.1 claims a 1.08% mAP improvement over RDMNet on VRain, but Table I gives RDMNet 78.41 and UniDet-D 79.10, a difference of 0.69%. Section IV.C.2 claims gains of 0.95% over TogetherNet and 1.32% over RDMNet on VSnow, but Table II gives differences of 1.32% and 0.95%, respectively (the pair is swapped). Since the central contribution is a small, consistent mAP advantage, the prose must exactly agree with the tabulated values.
  2. [III.B.3, Eqs. (11)–(12), and IV.C.5] The generalization story depends on learned binary frequency switches (AF2) trained under synthetic rain, fog, and snow transferring to unseen degradation types, but the paper never verifies which frequencies are selected or that those selections transfer. Table VI ablates the switches only on VFog and reports only mAP, so it cannot separate AF2's contribution from the MSP projection or the multi-encoder fusion. The sole quantitative unseen test is ExDark (low-light), where gains could come from the restoration branch or the multi-encoder design rather than from frequency switching, and sandstorm and rain-fog results are visual only. The sentence at the end of IV.D.2 attributing the generalization 'largely' to dynamic spectral learning is therefore unsupported without a direct analysis of switch patterns.
  3. [IV.C.5 and Table V] The quantitative generalization evidence on ExDark lacks error bars or seed variance, and the margin over RDMNet is only 1.85 mAP in a zero-shot setting. With a single run and no statistical significance assessment, the claim of superior generalization to unseen real-world degradation is fragile. The authors should report results over multiple seeds or provide a variance estimate, and ideally include an ablation on ExDark with AF2 disabled to attribute the gain to the proposed mechanism.
minor comments (4)
  1. [I and V] The phrase 'a AF2 module' appears in the Introduction and Conclusion; it should be 'an AF2 module.'
  2. [I] The contribution bullet and the abstract state 'cross diverse degradation types'; the standard phrasing is 'across diverse degradation types.'
  3. [Fig. 3] The figure caption says 'DCT Bases' but the surrounding text refers to 'DCT basis functions'; the notation and caption should be harmonized for clarity.
  4. [Eq. (7)] The reconstruction in Eq. (7) repeats the DC term from Eq. (6) with the same coefficient; the derivation is correct but the presentation would benefit from a brief note that the basis is not orthonormalized, since the constant factor H·W depends on the definition in Eq. (4).

Circularity Check

0 steps flagged · score 2.0 of 10

No significant circularity: UniDet-D is benchmarked against external baselines and its learned spectral switches are a fitted component, not a prediction defined by its own outputs.

full rationale

No load-bearing circular step can be exhibited. The paper's central comparisons (Tables I-V) are against externally trained or published baselines (YOLOXs, TogetherNet, RDMNet, AirNet/RestorNet pipelines, etc.) on separate test sets, so the performance claims are not defined in terms of the model's own fitted values. The AF2 switch mechanism is learned end-to-end by Gumbel-Softmax sampling with a sparsity regularizer (Eqs. 11-13); it is a trained component rather than a 'prediction' derived from the benchmark. The DCT analysis (Eqs. 3-7) is a standard mathematical identity showing GAP equals the DC component of the DCT; although it motivates MSP, it does not presuppose the detection result. The only self-citations ([5], [6]) are remote-sensing works cited in a general autonomy sentence and are not load-bearing. The generalization claim to unseen sandstorms/rain-fog is supported only qualitatively and the selected frequency patterns are not directly inspected, but that is an evidence gap, not circularity, because the metric is externally benchmarked and not definitionally tied to the module's outputs.

Assumptions & free parameters 6 free parameters · 5 assumptions · 2 invented entities

The central claim rests on a small number of premises the reader does not pay for: the known GAP-equals-DC observation that motivates frequency attention, the assumption that learned frequency gating transfers to unseen degradations, and the fidelity of synthetic weather data. The loss weights are borrowed from prior work, but the segment count n, the per-segment frequency assignment, and the Gumbel temperature are unstated design choices. No new physical entities are introduced; MSP and AF2 are architectural components whose independent evidence is limited to the paper's own ablations.

free parameters (6)
  • Number of DCT segments n (MSP channel partition)
    The paper defines C' = C/n but never states n, a hand-chosen architectural hyperparameter controlling how many frequency components are used. Section III.B.2.
  • Per-segment DCT frequency index (u_i, v_i) = not stated
    Each channel segment gets a designated frequency component; the assignment rule is unspecified. Section III.B.2, Eq. (9).
  • Loss weights lambda1-lambda4 = 0.2, 0.8, 0.1, 0.1
    Set 'following previous works' [40,42,43]; borrowed tuned constants balancing detection, restoration, frequency sparsity, and contrastive terms. Eq. (13).
  • Gumbel-Softmax temperature
    Required for the binary switch sampling in AF2 but never specified in the paper. Section III.B.3.
  • Low-light gamma exponent range r = [1.5, 5]
    Data-synthesis parameter for VLowLight; the chosen range shapes the low-light training distribution. Section IV.B.
  • Snow mask intensity weights = 0.5 to 1.0
    Data-synthesis parameter for VSnow. Section IV.B.
assumptions (5)
  • standard math Inverse 2D DCT reconstruction of a feature map from its frequency coefficients
    The motivating analysis relies on DCT orthogonality; as written, Eq. (6) omits the normalization factors of the non-orthonormal DCT-II basis, though the GAP-equals-DC conclusion of Eq. (5) is standard.
  • domain assumption CNNs are biased toward low-frequency signals and underuse high-frequency detail
    Borrowed from FCANet [19] and related work [20]; it motivates MSP and AF2 but is asserted rather than established in this paper. Section III.B.1.
  • domain assumption Learned binary frequency selection transfers across degradation types
    The core mechanism assumption; supported only by a two-row ablation (Table VI) and by qualitative examples for unseen weather. Section III.B.3, IV.C.5.
  • domain assumption Synthetic degradations faithfully represent real adverse weather
    Training and most test data are synthetic (RainDS rain, scattering-model fog, CSD snow, gamma low-light); the only real-world test is ExDark. Section IV.B.
  • domain assumption The joint restoration branch aids detection
    Restoration output is never evaluated and is not ablated, so its contribution is assumed. Section III.A, IV.
invented entities (2)
  • Multi-Spectrum Perception (MSP) module
    purpose: Computes per-segment DCT projections to replace global average pooling in channel attention
    Architectural module; its benefit appears only in the paper's own ablation, with no external benchmark or formal analysis.
  • Adaptable Frequency Filtering (AF2) module
    purpose: Learns binary on/off switches over DCT frequency components via Gumbel-Softmax sampling
    Architectural module whose adaptive behavior is evidenced only by end-task mAP and qualitative figures within this paper.

how reviews work

0 comments
Cite this review

Pith. "Pith review of UniDet-D: A Unified Dynamic Spectral Attention Model for Object Detection under Adverse Weathers." pith.science (2026). https://pith.science/paper/UATCJ3WI

@misc{pith2026250612324,
  author       = {Pith},
  title        = {Pith review of: UniDet-D: A Unified Dynamic Spectral Attention Model for Object Detection under Adverse Weathers},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/UATCJ3WI}},
  note         = {Machine review of arXiv:2506.12324}
}
read the original abstract

Real-world object detection is a challenging task where the captured images/videos often suffer from complex degradations due to various adverse weather conditions such as rain, fog, snow, low-light, etc. Despite extensive prior efforts, most existing methods are designed for one specific type of adverse weather with constraints of poor generalization, under-utilization of visual features while handling various image degradations. Leveraging a theoretical analysis on how critical visual details are lost in adverse-weather images, we design UniDet-D, a unified framework that tackles the challenge of object detection under various adverse weather conditions, and achieves object detection and image restoration within a single network. Specifically, the proposed UniDet-D incorporates a dynamic spectral attention mechanism that adaptively emphasizes informative spectral components while suppressing irrelevant ones, enabling more robust and discriminative feature representation across various degradation types. Extensive experiments show that UniDet-D achieves superior detection accuracy across different types of adverse-weather degradation. Furthermore, UniDet-D demonstrates superior generalization towards unseen adverse weather conditions such as sandstorms and rain-fog mixtures, highlighting its great potential for real-world deployment.

Figures

Figures reproduced from arXiv: 2506.12324 by the authors.

Figure 1
Figure 1. The proposed UniDet-D is a unified end-to-end model capable of [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. (a) Overall architecture of the proposed UniDet-D framework, which integrates image restoration and object detection in a unified manner under [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. A more detailed visualization of the adaptable frequency filtering [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figures from the paper (2 more)
Figure 4
Figure 4. Figure 4: Visualization of detection results under different weather degradations (synthetic and real-world). UniDet-D consistently achieves more accurate [PITH_FULL_IMAGE:figures/full_fig_p008_4.png]
Figure 5
Figure 5. Figure 5: Visualization of detection results under unseen weather degradations, [PITH_FULL_IMAGE:figures/full_fig_p009_5.png]

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

56 extracted references · 51 canonical work pages

  1. [1]

    Yolox: Exceeding yolo series in 2021,

    Z. Ge, S. Liu, F. Wang et al. , “Yolox: Exceeding yolo series in 2021,” arXiv preprint arXiv:2107.08430 , 2021

    arXiv 2021

  2. [2]

    Perception and sensing for autonomous vehicles under adverse weather conditions: A survey,

    Y . Zhang, A. Carballo, H. Yang et al. , “Perception and sensing for autonomous vehicles under adverse weather conditions: A survey,” ISPRS J. Photogramm. Remote Sens. , vol. 196, pp. 146–177, 2023

    work page 2023

  3. [3]

    Learning to remove bad weather: Towards robust visual perception for self-driving,

    Y . Lee, Y . Kim, J. Yuet al., “Learning to remove bad weather: Towards robust visual perception for self-driving,” IEEE Rob. Autom. Lett. , pp. 1–1, 2022

    work page 2022

  4. [4]

    Object detection in 20 years: A survey,

    Z. Zou, K. Chen, Z. Shi et al., “Object detection in 20 years: A survey,” P IEEE , vol. 111, no. 3, pp. 257–276, 2023

    work page 2023

  5. [5]

    Earth- Marker: A visual prompting multi-modal large language model for remote sensing,

    W. Zhang, M. Cai, T. Zhang, Y . Zhuang, J. Li, and X. Mao, “Earth- Marker: A visual prompting multi-modal large language model for remote sensing,” IEEE Trans. Geosci. Remote Sens. , vol. 63, pp. 1–19, Jan. 2025

    work page 2025

  6. [6]

    EarthGPT: A universal multi-modal large language model for multi-sensor image comprehension in remote sensing domain,

    W. Zhang, M. Cai, T. Zhang, Y . Zhuang, and X. Mao, “EarthGPT: A universal multi-modal large language model for multi-sensor image comprehension in remote sensing domain,” IEEE Trans. Geosci. Remote Sens., vol. 62, pp. 1–20, Jun. 2024

    work page 2024

  7. [7]

    Adamixer: A fast-converging query- based object detector,

    Z. Gao, L. Wang, B. Han et al. , “Adamixer: A fast-converging query- based object detector,” in Proc. IEEE/CVF Conf. Comput. Vis. Pattern Recognit., June 2022, pp. 5364–5373. 10

    work page 2022

  8. [8]

    Image-adaptive yolo for object detection in adverse weather conditions,

    W. Liu, G. Ren, R. Yu et al., “Image-adaptive yolo for object detection in adverse weather conditions,” in AAAI Conf. Artif. Intell. , vol. 36, no. 2, 2022, pp. 1792–1800

    work page 2022

Show all 56 references
  1. [9]

    Improving nighttime driving-scene segmentation via dual image-adaptive learnable filters,

    W. Liu, W. Li, J. Zhu et al. , “Improving nighttime driving-scene segmentation via dual image-adaptive learnable filters,” IEEE Trans. Circuits Syst. Video Technol. , vol. 33, no. 10, pp. 5855–5867, 2023

    2023

  2. [10]

    Gdip: Gated differentiable image processing for object detection in adverse conditions,

    S. Kalwar, D. Patel, A. Aanegola et al. , “Gdip: Gated differentiable image processing for object detection in adverse conditions,” in IEEE Int. Conf. Robot. Autom. , 2023, pp. 7083–7089

    2023

  3. [11]

    Detection-driven exposure-correction network for nighttime drone-view object detection,

    Y . Xi, W. Jia, Q. Miao et al. , “Detection-driven exposure-correction network for nighttime drone-view object detection,” IEEE Trans. Geosci. Remote Sens. , 2024

    2024

  4. [12]

    Cpa-enhancer: Chain-of-thought prompted adaptive enhancer for object detection under unknown degra- dations,

    Y . Zhang, Y . Wu, Y . Liu et al. , “Cpa-enhancer: Chain-of-thought prompted adaptive enhancer for object detection under unknown degra- dations,” arXiv preprint arXiv:2403.11220 , 2024

    2024 arXiv

  5. [13]

    Unsupervised domain adaptation of object detectors: A survey,

    P. Oza, V . A. Sindagi, V . VS et al. , “Unsupervised domain adaptation of object detectors: A survey,” IEEE Trans. Pattern Anal. Mach. Intell. , vol. 46, no. 6, pp. 4018–4040, 2024

    2024

  6. [14]

    Cross-domain object detection for autonomous driving: A stepwise domain adaptative yolo approach,

    G. Li, Z. Ji, X. Qu et al., “Cross-domain object detection for autonomous driving: A stepwise domain adaptative yolo approach,” IEEE Trans. Intell. V eh., vol. 7, no. 3, pp. 603–615, 2022

    2022

  7. [15]

    Togethernet: Bridging image restoration and object detection together via dynamic enhancement learning,

    Y . Wang, X. Yan, K. Zhang et al. , “Togethernet: Bridging image restoration and object detection together via dynamic enhancement learning,” Comput. Graphics F orum, vol. 41, no. 7, pp. 465–476, 2022

    2022

  8. [16]

    Dsnet: Joint semantic learning for object detection in inclement weather conditions,

    S.-C. Huang, T.-H. Le, and D.-W. Jaw, “Dsnet: Joint semantic learning for object detection in inclement weather conditions,” IEEE Trans. Pattern Anal. Mach. Intell. , vol. 43, no. 8, pp. 2623–2633, 2021

    2021

  9. [17]

    Dehazing & reasoning yolo: Prior knowledge-guided network for object detection in foggy weather,

    F. Zhong, W. Shen, H. Yu et al. , “Dehazing & reasoning yolo: Prior knowledge-guided network for object detection in foggy weather,” Pattern Recognit., vol. 156, p. 110756, 2024

    2024

  10. [18]

    Mdd-shipnet: Math-data integrated defogging for fog-occlusion ship detection,

    N. Wang, Y . Wang, Y . Feng et al. , “Mdd-shipnet: Math-data integrated defogging for fog-occlusion ship detection,” IEEE Trans. Intell. Transp. Syst., 2024

    2024

  11. [19]

    Fcanet: Frequency channel attention networks,

    Z. Qin, P. Zhang, F. Wu et al. , “Fcanet: Frequency channel attention networks,” in Proc. IEEE/CVF Int. Conf. Comput. Vis. , 2021, pp. 783– 792

    2021

  12. [20]

    Boosting night-time scene parsing with learnable frequency,

    Z. Xie, S. Wang, K. Xu et al. , “Boosting night-time scene parsing with learnable frequency,” IEEE Trans. Image Process. , vol. 32, pp. 2386– 2398, 2023

    2023

  13. [21]

    Discrete cosine transform,

    N. Ahmed, T. Natarajan, and K. R. Rao, “Discrete cosine transform,” IEEE Trans. Comput. , vol. 100, no. 1, pp. 90–93, 1974

    1974

  14. [22]

    Libra r-cnn: Towards balanced learning for object detection,

    J. Pang, K. Chen, J. Shi et al., “Libra r-cnn: Towards balanced learning for object detection,” in Proc. IEEE Conf. Comput. Vis. Pattern Recog. , 2019, pp. 821–830

    2019

  15. [23]

    Dynamic r-cnn: Towards high quality object detection via dynamic training,

    H. Zhang, H. Chang, B. Ma et al., “Dynamic r-cnn: Towards high quality object detection via dynamic training,” in Proc. Eur . Conf. Comput. Vis. Springer, 2020, pp. 260–275

    2020

  16. [24]

    Selective search for object recognition,

    J. R. Uijlings, K. E. Van De Sande et al. , “Selective search for object recognition,” Int. J. Comput. Vis. , vol. 104, pp. 154–171, 2013

    2013

  17. [25]

    Ssd: Single shot multibox detector,

    W. Liu, D. Anguelov, D. Erhan et al. , “Ssd: Single shot multibox detector,” in Proc. Eur . Conf. Comput. Vis. Springer, 2016, pp. 21– 37

    2016

  18. [26]

    Focal loss for dense object detection,

    T.-Y . Lin, P. Goyal, R. Girshick et al. , “Focal loss for dense object detection,” in IEEE Conf. Comput. Vis. Pattern Recognit. , 2017, pp. 2980–2988

    2017

  19. [27]

    Objects as points,

    X. Zhou, D. Wang, and P. Kr ¨ahenb¨uhl, “Objects as points,” arXiv preprint arXiv:1904.07850, 2019

    1904 arXiv

  20. [28]

    Yolov7: Trainable bag-of-freebies sets new state-of-the-art for real-time object detectors,

    C.-Y . Wang, A. Bochkovskiy, and H.-Y . M. Liao, “Yolov7: Trainable bag-of-freebies sets new state-of-the-art for real-time object detectors,” in Proc. IEEE Conf. Comput. Vis. Pattern Recog. (CVPR) , June 2023, pp. 7464–7475

    2023

  21. [29]

    Yolov9: Learning what you want to learn using programmable gradient information,

    C.-Y . Wang, I.-H. Yeh, and H.-Y . Mark Liao, “Yolov9: Learning what you want to learn using programmable gradient information,” in Proc. Eur . Conf. Comput. Vis. Springer, 2025, pp. 1–21

    2025

  22. [30]

    Yolov10: Real-time end-to-end object detection,

    A. Wang, H. Chen, L. Liu et al., “Yolov10: Real-time end-to-end object detection,” arXiv preprint arXiv:2405.14458 , 2024

    2024 arXiv

  23. [31]

    Spectral unsupervised domain adaptation for visual recognition,

    J. Zhang, J. Huang, Z. Tian et al. , “Spectral unsupervised domain adaptation for visual recognition,” in Proc. IEEE Conf. Comput. Vis. Pattern Recog., 2022, pp. 9829–9840

    2022

  24. [32]

    Domain adaptive object detection for autonomous driving under foggy weather,

    J. Li, R. Xu, J. Ma et al. , “Domain adaptive object detection for autonomous driving under foggy weather,” in Proc. IEEE Winter Conf. Appl. Comput. Vis. , January 2023, pp. 612–622

    2023

  25. [33]

    2pcnet: Two-phase consistency training for day-to-night unsupervised domain adaptive object detection,

    M. Kennerley, J.-G. Wang, B. Veeravalli, and R. T. Tan, “2pcnet: Two-phase consistency training for day-to-night unsupervised domain adaptive object detection,” in Proc. IEEE/CVF Conf. Comput. Vis. Pattern Recognit., June 2023, pp. 11 484–11 493

    2023

  26. [34]

    R-yolo: A robust object detector in adverse weather,

    L. Wang, H. Qin, X. Zhou et al. , “R-yolo: A robust object detector in adverse weather,” IEEE Trans. Instrum. Meas. , vol. 72, pp. 1–11, 2023

    2023

  27. [35]

    Mcnet: Magnitude consistency network for domain adaptive object detection under inclement environments,

    J. Pang, “Mcnet: Magnitude consistency network for domain adaptive object detection under inclement environments,” Pattern Recognit., vol. 145, p. 109947, 2024

    2024

  28. [36]

    Fsdr: Frequency space domain randomization for domain generalization,

    J. Huang, D. Guan, A. Xiao, and S. Lu, “Fsdr: Frequency space domain randomization for domain generalization,” in Proc. IEEE Conf. Comput. Vis. Pattern Recog., 2021, pp. 6891–6902

    2021

  29. [37]

    Multitask aet with orthogonal tangent regularity for dark object detection,

    Z. Cui, G.-J. Qi, L. Gu et al. , “Multitask aet with orthogonal tangent regularity for dark object detection,” in Proceedings of the Proc. IEEE Int. Conf. Comput. Vis. , October 2021, pp. 2553–2562

    2021

  30. [38]

    Coderainnet: Collaborative deraining network for drone-view object detection in rainy weather conditions,

    Y . Xi, W. Jia, Q. Miao et al. , “Coderainnet: Collaborative deraining network for drone-view object detection in rainy weather conditions,” Remote Sens. , vol. 15, no. 6, p. 1487, 2023

    2023

  31. [39]

    3d semantic segmentation in the wild: Learning generalized models for adverse-condition point clouds,

    A. Xiao, J. Huang, W. Xuan, R. Ren, K. Liu, D. Guan, A. El Saddik, S. Lu, and E. P. Xing, “3d semantic segmentation in the wild: Learning generalized models for adverse-condition point clouds,” in Proc. IEEE Conf. Comput. Vis. Pattern Recog. , 2023, pp. 9382–9392

    2023

  32. [40]

    Degradation modeling for restoration- enhanced object detection in adverse weather scenes,

    X. Wang, X. Liu, H. Yang et al., “Degradation modeling for restoration- enhanced object detection in adverse weather scenes,” IEEE Trans. Intell. V eh., 2024

    2024

  33. [41]

    Categorical reparametrization with gumble-softmax,

    E. Jang, S. Gu, and B. Poole, “Categorical reparametrization with gumble-softmax,” in Proc. Int. Conf. Learn. Represent. OpenReview. net, 2017

    2017

  34. [42]

    Frequency learning attention networks based on deep learning for automatic modulation classification in wireless communication,

    D. Zhang, Y . Lu, Y . Li et al. , “Frequency learning attention networks based on deep learning for automatic modulation classification in wireless communication,” Pattern Recognit., vol. 137, p. 109345, 2023

    2023

  35. [43]

    Restornet: An efficient network for multiple degradation image restoration,

    X. Wang, H. Chen, H. Gou et al. , “Restornet: An efficient network for multiple degradation image restoration,” Knowledge-Based Systems, vol. 282, p. 111116, 2023

    2023

  36. [44]

    All-in-one image restoration for unknown corruption,

    B. Li, X. Liu, P. Hu et al. , “All-in-one image restoration for unknown corruption,” in Proc. IEEE Conf. Comput. Vis. Pattern Recog. , 2022, pp. 17 452–17 462

    2022

  37. [45]

    Learning multiple adverse weather removal via two-stage knowledge learning and multi-contrastive regularization: Toward a unified model,

    W.-T. Chen, Z.-K. Huang, C.-C. Tsai et al., “Learning multiple adverse weather removal via two-stage knowledge learning and multi-contrastive regularization: Toward a unified model,” in Proc. IEEE Conf. Comput. Vis. Pattern Recog., 2022, pp. 17 653–17 662

    2022

  38. [46]

    Lmqformer: A laplace-prior-guided mask query transformer for lightweight snow removal,

    J. Lin, N. Jiang, Z. Zhang et al. , “Lmqformer: A laplace-prior-guided mask query transformer for lightweight snow removal,” IEEE Trans. Circuits Syst. Video Technol. , vol. 33, no. 11, pp. 6225–6235, 2023

    2023

  39. [47]

    Single image haze removal using dark channel prior,

    K. He, J. Sun, and X. Tang, “Single image haze removal using dark channel prior,” IEEE Trans. Pattern Anal. Mach. Intell. , vol. 33, no. 12, pp. 2341–2353, 2010

    2010

  40. [48]

    Aod-net: All-in-one dehazing network,

    B. Li, X. Peng, Z. Wang et al., “Aod-net: All-in-one dehazing network,” in IEEE Conf. Comput. Vis. Pattern Recognit. , 2017, pp. 4770–4778

    2017

  41. [49]

    Contrastive learning for compact single image dehazing,

    H. Wu, Y . Qu, S. Lin et al. , “Contrastive learning for compact single image dehazing,” in Proc. IEEE Conf. Comput. Vis. Pattern Recog. , 2021, pp. 10 551–10 560

    2021

  42. [50]

    Zero-reference deep curve estimation for low-light image enhancement,

    C. Guo et al., “Zero-reference deep curve estimation for low-light image enhancement,” in Proc. IEEE Conf. Comput. Vis. Pattern Recog. , 2020, pp. 1780–1789

    2020

  43. [51]

    You only need 90k parameters to adapt light: a light weight transformer for image enhancement and exposure correction,

    Z. Cui et al. , “You only need 90k parameters to adapt light: a light weight transformer for image enhancement and exposure correction,” arXiv preprint arXiv:2205.14871 , 2022

    2022 arXiv

  44. [52]

    The pascal visual object classes (voc) challenge,

    M. Everingham, L. Van Gool, C. K. Williams et al., “The pascal visual object classes (voc) challenge,” Int. J. Comput. Vis. , vol. 88, pp. 303– 338, 2010

    2010

  45. [53]

    Removing raindrops and rain streaks in one go,

    R. Quan, X. Yu, Y . Liang et al. , “Removing raindrops and rain streaks in one go,” in Proc. IEEE Conf. Comput. Vis. Pattern Recog. , 2021, pp. 9147–9156

    2021

  46. [54]

    All snow removed: Single image desnowing algo- rithm using hierarchical dual-tree complex wavelet representation and contradict channel loss,

    W.-T. Chen et al. , “All snow removed: Single image desnowing algo- rithm using hierarchical dual-tree complex wavelet representation and contradict channel loss,” in Proc. IEEE/CVF Int. Conf. Comput. Vis. , 2021, pp. 4196–4205

    2021

  47. [55]

    Benchmarking single-image dehazing and beyond,

    B. Li, W. Ren, D. Fu et al. , “Benchmarking single-image dehazing and beyond,” IEEE Trans. Image Process., vol. 28, no. 1, pp. 492–505, 2018

    2018

  48. [56]

    Getting to know low-light images with the exclusively dark dataset,

    Y . P. Loh and C. S. Chan, “Getting to know low-light images with the exclusively dark dataset,” Comput. Vis. Image Underst. , vol. 178, pp. 30–42, 2019

    2019

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.