pith. sign in

arxiv: 2606.11186 · v1 · pith:T7XZKOHCnew · submitted 2026-06-09 · 💻 cs.CV

AnyMod-LLVE: Low-Light Video Enhancement with Modality-Agnostic Inference

Hangfeng Liang , Yutao Hu , Yanhan Hu , Xiaohan Wu , Wenqi Shao , Ying Fu This is my paper

Pith reviewed 2026-06-27 13:32 UTC · model grok-4.3

classification 💻 cs.CV
keywords low-light video enhancementmultimodal inferencemodality-agnosticevent streamsinfrared imagessynthetic pretrainingcross-modal translatorimplicit representations
0
0 comments X

The pith

AMNet performs low-light video enhancement with any combination of modalities available at inference by generating implicit auxiliary representations from synthetic pretraining.

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

The paper presents AMNet as a unified framework for low-light video enhancement that does not require auxiliary modalities such as event streams or infrared images to be present during inference. It achieves this through a Spatial-Spectral Dual-Gated Translator trained via large-scale pretraining on RGB-only data augmented with synthetic auxiliaries, which produces implicit representations to substitute for missing inputs. This design enables flexible handling of arbitrary modality combinations while delivering strong enhancement performance even under complete modality absence. The work targets the practical gap where auxiliary sensors cannot be guaranteed in real deployments.

Core claim

AMNet is a unified multimodal framework for LLVE that supports flexible modality-agnostic inference. To address modality absence, it introduces a Spatial-Spectral Dual-Gated Translator that learns the correspondence between auxiliary modalities and RGB inputs from synthetic auxiliary modalities during large-scale multimodal pretraining on RGB-only datasets, thereby producing implicit auxiliary representations to support robust enhancement when real auxiliary modalities are unavailable at inference.

What carries the argument

Spatial-Spectral Dual-Gated Translator, which learns cross-modal correspondences from synthetic auxiliaries to produce implicit auxiliary representations that substitute for absent real modalities during inference.

Load-bearing premise

Correspondences learned from synthetic auxiliary modalities in pretraining will generalize to generate useful implicit representations when real auxiliary modalities are missing at inference.

What would settle it

A controlled test where the pretraining stage with synthetic auxiliaries is removed and performance on real data with absent modalities falls to the level of standard RGB-only baselines.

Figures

Figures reproduced from arXiv: 2606.11186 by Hangfeng Liang, Wenqi Shao, Xiaohan Wu, Yanhan Hu, Ying Fu, Yutao Hu.

Figure 1
Figure 1. Figure 1: Comparison of different LLVE paradigms under missing￾modality scenarios. (a) RGB-based LLVE relies solely on de￾graded RGB inputs and struggles to recover fine details. (b) Multi￾modal LLVE improves enhancement quality by leveraging auxil￾iary modalities (e.g., events), but fails when such modalities are missing at inference time. (c) Modality-agnostic LLVE enables robust enhancement by exploiting auxiliar… view at source ↗
Figure 2
Figure 2. Figure 2: Overview architecture of the proposed AMNet. Notably, auxiliary modalities are optional, and the network remains robust under modality absence scenario via the synthetic implicit representation from S2DG Translator. FFT FBS ⨁ iFFT Conv2D Conv2D Spectral Gating Conv 2 D Sigmoid Conv2D Spectral Scaling Conv2D Conv2D tanh IADS 𝑴𝒔𝒑𝒂𝒕𝒊𝒂𝒍 ⊖ 𝒁𝒍𝒐𝒘 𝒁𝒉𝒊𝒈𝒉 Conv2D Conv2D Sigmoid 𝒁𝒕 𝒓𝒈𝒃 AvgPool 𝒁 ෡ 𝒕 𝒎 𝒁 ෩ 𝒉𝒊𝒈𝒉 𝒁 ෡ 𝒉𝒊𝒈… view at source ↗
Figure 3
Figure 3. Figure 3: The details of the proposed S2DG Translator. The RGB features and the auxiliary modality features are then fused and fed into a temporal modeling module to cap￾ture inter-frame dependencies. Finally, a decoder predicts a residual map, which will be combined with low-light input Rlow t to form the final output Ren t . 3.2. Spatial-Spectral Dual-Gated Translator Auxiliary modalities such as event streams and… view at source ↗
Figure 4
Figure 4. Figure 4: Visual comparison on RGB only dataset (DID). Notably, in this experiment, our AMNet only takes RGB as input without the auxiliary modality. public video datasets as sources for pretraining, covering diverse scene types, motion patterns, and content distri￾butions. The statistics of these datasets are summarized in [PITH_FULL_IMAGE:figures/full_fig_p006_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Visual comparison with multimodal LLVE methods on the SDE dataset under different modality availability. When auxiliary modalities are absent at inference (R), existing multimodal methods (e.g., EvLight++) exhibit noticeable performance degradation. In contrast, AMNet maintains stable enhancement quality and preserves fine-grained details [PITH_FULL_IMAGE:figures/full_fig_p007_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Visualization of real and synthetic feature representation for auxiliary modalities feature maps. the learning of cross-modal correspondence, bring more realistic synthesized representations for missing modality and thereby improves the performance. Ablation on Model Components. We investigate the con￾tribution of the key components in the proposed Spatial￾Spectral Dual-Gated (S2DG) Translator [PITH_FULL_… view at source ↗
Figure 7
Figure 7. Figure 7: Visualization of spatial filtering and frequency-band gating in the S2DG Translator [PITH_FULL_IMAGE:figures/full_fig_p014_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: Feature distribution comparison between real and synthesized auxiliary modalities. Effect of Training Modalities. We first evaluate how auxiliary modalities contribute during training when only RGB inputs are available at inference time (Table 8A). Compared with RGB-only training, introducing event or infrared supervision consistently improves performance across all datasets. In particular, infrared modali… view at source ↗
Figure 9
Figure 9. Figure 9: Qualitative comparison of enhancement results from AMNet using different auxiliary modalities. Input Low-Light SGZ Zero-TIG FoundIR MobileIE DarkIR AMNet (Ours) GT [PITH_FULL_IMAGE:figures/full_fig_p016_9.png] view at source ↗
Figure 10
Figure 10. Figure 10: Qualitative comparison of zero-shot low-light video enhancement on unseen datasets. Without any finetuning, AMNet consistently recovers clearer structures and finer details than representative restoration foundation models, producing results closer to the ground truth. 16 [PITH_FULL_IMAGE:figures/full_fig_p016_10.png] view at source ↗
read the original abstract

Low-light video enhancement (LLVE) remains a challenging task due to severe information degradation under low-illumination conditions. Recent multimodal approaches have significantly improved enhancement performance by incorporating auxiliary modalities, such as event streams and infrared images. However, these methods typically assume the availability of these modalities at inference, which is often not feasible in real-world scenarios. To solve this problem, in this work, we propose AMNet, a unified multimodal framework for LLVE, to support flexible modality-agnostic inference, where auxiliary modalities may be unavailable. To address the issue of modality absence, we introduce a Spatial-Spectral Dual-Gated Translator that learns the correspondence between auxiliary modalities and RGB inputs, producing implicit auxiliary representations to support the robust enhancement. Additionally, to fully facilitate the learning of cross-modal correspondence, we conduct large-scale multimodal pretraining based on the RGB-only dataset with synthetic auxiliary modalities. Extensive experiments demonstrate that AMNet could handle arbitrary inference-time modality combinations and exhibits superior performance for LLVE under modality absence conditions. Code and models are available on the project page.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 1 minor

Summary. The paper proposes AMNet, a unified multimodal framework for low-light video enhancement (LLVE) supporting modality-agnostic inference at test time. It introduces a Spatial-Spectral Dual-Gated Translator pretrained on large-scale synthetic auxiliary modalities (generated from RGB-only data) to produce implicit representations that substitute for absent real auxiliaries such as event streams or infrared images. The central claim is that this enables handling arbitrary inference-time modality combinations while achieving superior LLVE performance under modality absence conditions.

Significance. If the claims hold, the result would be significant for practical LLVE deployment, as it removes the requirement for auxiliary sensors at inference while retaining benefits from multimodal pretraining. The pretraining strategy on synthetic data from RGB-only sources is a clear strength, as it addresses the scarcity of real multimodal low-light video datasets.

major comments (2)
  1. [Abstract and method description of Spatial-Spectral Dual-Gated Translator] The load-bearing claim that the Spatial-Spectral Dual-Gated Translator generalizes from synthetic auxiliaries (used exclusively in pretraining) to produce useful implicit representations when real auxiliaries are absent at inference lacks supporting evidence. No ablations isolating this transfer (e.g., performance with real vs. synthetic auxiliary inputs at test time, or controls for noise/temporal/spectral mismatch) are described.
  2. [Abstract] The abstract asserts superior performance and extensive experiments but supplies no quantitative results, baselines, ablation details, or dataset information. Without these, the extent of improvement under modality absence conditions cannot be evaluated against prior multimodal LLVE methods.
minor comments (1)
  1. [Method] Clarify the precise integration of the Spatial-Spectral Dual-Gated Translator output into the main enhancement backbone and any additional loss terms used during pretraining.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive feedback on our manuscript. We address each major comment below and will revise the paper to strengthen the presentation of evidence and clarity of claims.

read point-by-point responses

  1. Referee: [Abstract and method description of Spatial-Spectral Dual-Gated Translator] The load-bearing claim that the Spatial-Spectral Dual-Gated Translator generalizes from synthetic auxiliaries (used exclusively in pretraining) to produce useful implicit representations when real auxiliaries are absent at inference lacks supporting evidence. No ablations isolating this transfer (e.g., performance with real vs. synthetic auxiliary inputs at test time, or controls for noise/temporal/spectral mismatch) are described.

    Authors: We agree that explicit ablations isolating the synthetic-to-implicit transfer would strengthen the evidence. The current experiments (Sections 4.2 and 4.3) demonstrate that AMNet outperforms baselines under complete modality absence at inference, which relies on the implicit representations produced by the translator after synthetic pretraining. However, real multimodal low-light video datasets with aligned event/IR streams are not publicly available at scale, which is the motivation for our synthetic auxiliary generation approach. We will add a dedicated ablation subsection discussing domain gap, including controls for temporal alignment and spectral mismatch where synthetic perturbations can be applied, and clarify the generalization argument. revision: yes

  2. Referee: [Abstract] The abstract asserts superior performance and extensive experiments but supplies no quantitative results, baselines, ablation details, or dataset information. Without these, the extent of improvement under modality absence conditions cannot be evaluated against prior multimodal LLVE methods.

    Authors: We acknowledge that the abstract is currently qualitative. We will revise it to include key quantitative highlights (e.g., average PSNR/SSIM gains on representative datasets under modality-absent settings), the number of baselines compared, and the scale of pretraining data, while remaining within length constraints. revision: yes

Circularity Check

0 steps flagged

No significant circularity detected

full rationale

The paper's derivation relies on a new architectural component (Spatial-Spectral Dual-Gated Translator) and a pretraining procedure that generates synthetic auxiliaries from RGB data to learn cross-modal correspondences. These are presented as independent design choices whose effectiveness is evaluated empirically on modality-absence scenarios. No equations or claims reduce a prediction to a fitted parameter by construction, no load-bearing self-citations close the central argument, and no uniqueness theorems or ansatzes are imported from prior author work. The generalization step from synthetic to real absent modalities is an empirical hypothesis, not a definitional identity.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 1 invented entities

The central claim depends on the domain assumption that synthetic auxiliary data can train effective cross-modal mappings and on the new architectural component that produces implicit representations; no numerical free parameters are specified in the abstract.

axioms (1)
  • domain assumption Synthetic auxiliary modalities generated from RGB-only data can serve as effective proxies for training cross-modal correspondences that generalize to real missing modalities.
    Pretraining is performed on an RGB-only dataset augmented with synthetic auxiliary modalities.
invented entities (1)
  • Spatial-Spectral Dual-Gated Translator no independent evidence
    purpose: Learns correspondence between auxiliary modalities and RGB inputs to generate implicit auxiliary representations supporting enhancement when auxiliaries are absent.
    New module introduced specifically to enable modality-agnostic inference.

pith-pipeline@v0.9.1-grok · 5732 in / 1192 out tokens · 26049 ms · 2026-06-27T13:32:29.341220+00:00 · methodology

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

121 extracted references · 9 canonical work pages

  1. [1]

    Proceedings of the IEEE/CVF international conference on computer vision , pages=

    Retinexformer: One-stage retinex-based transformer for low-light image enhancement , author=. Proceedings of the IEEE/CVF international conference on computer vision , pages=

  2. [2]

    Proceedings of the IEEE/CVF International Conference on Computer Vision , pages=

    RetinexMCNet: A Memory Controller Dominated Network for Low-Light Video Enhancement Based on Retinex , author=. Proceedings of the IEEE/CVF International Conference on Computer Vision , pages=

  3. [3]

    European Conference on Computer Vision , pages=

    Towards real-world event-guided low-light video enhancement and deblurring , author=. European Conference on Computer Vision , pages=. 2024 , organization=

    2024

  4. [4]

    IEEE Transactions on Pattern Analysis and Machine Intelligence , year=

    Evlight++: Low-light video enhancement with an event camera: A large-scale real-world dataset, novel method, and more , author=. IEEE Transactions on Pattern Analysis and Machine Intelligence , year=

  5. [5]

    Pattern Recognition , volume=

    STARNet: Low-light video enhancement using spatio-temporal consistency aggregation , author=. Pattern Recognition , volume=. 2025 , publisher=

    2025

  6. [7]

    Proceedings of the AAAI Conference on Artificial Intelligence , volume=

    Thermal-Aware Low-Light Image Enhancement: A Real-World Benchmark and a New Light-Weight Model , author=. Proceedings of the AAAI Conference on Artificial Intelligence , volume=

  7. [8]

    Proceedings of the IEEE/CVF International conference on computer vision , pages=

    Seeing motion in the dark , author=. Proceedings of the IEEE/CVF International conference on computer vision , pages=

  8. [9]

    Proceedings of the IEEE/CVF international conference on computer vision , pages=

    Seeing dynamic scene in the dark: A high-quality video dataset with mechatronic alignment , author=. Proceedings of the IEEE/CVF international conference on computer vision , pages=

  9. [10]

    Proceedings of the IEEE/CVF International Conference on Computer Vision , pages=

    Dancing in the dark: A benchmark towards general low-light video enhancement , author=. Proceedings of the IEEE/CVF International Conference on Computer Vision , pages=

  10. [11]

    Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , pages=

    Learning temporal consistency for low light video enhancement from single images , author=. Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , pages=

  11. [13]

    Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages=

    Binarized low-light raw video enhancement , author=. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages=

  12. [15]

    Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , pages=

    Learning semantic-aware knowledge guidance for low-light image enhancement , author=. Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , pages=

  13. [16]

    Proceedings of the IEEE conference on computer vision and pattern recognition , pages=

    Learning to see in the dark , author=. Proceedings of the IEEE conference on computer vision and pattern recognition , pages=

  14. [17]

    European conference on computer vision , pages=

    Simple baselines for image restoration , author=. European conference on computer vision , pages=. 2022 , organization=

    2022

  15. [18]

    Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , pages=

    Restormer: Efficient transformer for high-resolution image restoration , author=. Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , pages=

  16. [19]

    Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , pages=

    Snr-aware low-light image enhancement , author=. Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , pages=

  17. [20]

    Proceedings of the Computer Vision and Pattern Recognition Conference , pages=

    Efficient Diffusion as Low Light Enhancer , author=. Proceedings of the Computer Vision and Pattern Recognition Conference , pages=

  18. [21]

    Computer Vision and Image Understanding , pages=

    AdaptDiff: Adaptive diffusion learning for low-light image enhancement , author=. Computer Vision and Image Understanding , pages=. 2025 , publisher=

    2025

  19. [22]

    2025 IEEE/CVF Winter Conference on Applications of Computer Vision (WACV) , pages=

    Event-guided low-light video semantic segmentation , author=. 2025 IEEE/CVF Winter Conference on Applications of Computer Vision (WACV) , pages=. 2025 , organization=

    2025

  20. [23]

    European Conference on Computer Vision , pages=

    Autodir: Automatic all-in-one image restoration with latent diffusion , author=. European Conference on Computer Vision , pages=. 2024 , organization=

    2024

  21. [24]

    Proceedings of the AAAI Conference on Artificial Intelligence , volume=

    Darkvisionnet: Low-light imaging via rgb-nir fusion with deep inconsistency prior , author=. Proceedings of the AAAI Conference on Artificial Intelligence , volume=

  22. [26]

    Advances in neural information processing systems , volume=

    Multimodal generative models for scalable weakly-supervised learning , author=. Advances in neural information processing systems , volume=

  23. [27]

    Proceedings of the IEEE/CVF International Conference on Computer Vision , pages=

    Unbiased missing-modality multimodal learning , author=. Proceedings of the IEEE/CVF International Conference on Computer Vision , pages=

  24. [28]

    International Journal of Multimedia Information Retrieval , volume=

    Chameleon: A Multimodal Learning Framework Robust to Missing Modalities , author=. International Journal of Multimedia Information Retrieval , volume=. 2025 , publisher=

    2025

  25. [30]

    International Conference on Medical Image Computing and Computer-Assisted Intervention , pages=

    Learnable cross-modal knowledge distillation for multi-modal learning with missing modality , author=. International Conference on Medical Image Computing and Computer-Assisted Intervention , pages=. 2023 , organization=

    2023

  26. [31]

    Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , pages=

    Learnable irrelevant modality dropout for multimodal action recognition on modality-specific annotated videos , author=. Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , pages=

  27. [33]

    Proceedings of the Computer Vision and Pattern Recognition Conference , pages=

    Resilient Sensor Fusion under Adverse Sensor Failures via Multi-Modal Expert Fusion , author=. Proceedings of the Computer Vision and Pattern Recognition Conference , pages=

  28. [34]

    Proceedings of the IEEE/CVF International Conference on Computer Vision , pages=

    Metabev: Solving sensor failures for 3d detection and map segmentation , author=. Proceedings of the IEEE/CVF International Conference on Computer Vision , pages=

  29. [35]

    Computer Vision and Image Understanding , volume=

    Cross-modal alignment and translation for missing modality action recognition , author=. Computer Vision and Image Understanding , volume=. 2023 , publisher=

    2023

  30. [36]

    Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , pages=

    Multi-modal learning with missing modality via shared-specific feature modelling , author=. Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , pages=

  31. [37]

    Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages=

    Events-to-video: Bringing modern computer vision to event cameras , author=. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages=

  32. [38]

    IEEE transactions on pattern analysis and machine intelligence , year=

    Frequency-aware feature fusion for dense image prediction , author=. IEEE transactions on pattern analysis and machine intelligence , year=

  33. [39]

    Proceedings of the IEEE/CVF International Conference on Computer Vision , pages=

    Frequency-dynamic attention modulation for dense prediction , author=. Proceedings of the IEEE/CVF International Conference on Computer Vision , pages=

  34. [40]

    Proceedings of the Computer Vision and Pattern Recognition Conference , pages=

    Frequency Dynamic Convolution for Dense Image Prediction , author=. Proceedings of the Computer Vision and Pattern Recognition Conference , pages=

  35. [41]

    Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages=

    A physics-based noise formation model for extreme low-light raw denoising , author=. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages=

  36. [42]

    International Journal of Computer Vision , volume=

    Video enhancement with task-oriented flow , author=. International Journal of Computer Vision , volume=. 2019 , publisher=

    2019

  37. [43]

    Proceedings of the IEEE/CVF conference on computer vision and pattern recognition workshops , pages=

    Ntire 2019 challenge on video deblurring and super-resolution: Dataset and study , author=. Proceedings of the IEEE/CVF conference on computer vision and pattern recognition workshops , pages=

    2019

  38. [44]

    Proceedings of the European conference on computer vision (ECCV) , pages=

    Youtube-vos: Sequence-to-sequence video object segmentation , author=. Proceedings of the European conference on computer vision (ECCV) , pages=

  39. [45]

    IEEE Transactions on Image Processing , volume=

    Adapool: Exponential adaptive pooling for information-retaining downsampling , author=. IEEE Transactions on Image Processing , volume=. 2022 , publisher=

    2022

  40. [46]

    Proceedings of the IEEE/CVF international conference on computer vision , pages=

    Unidentified video objects: A benchmark for dense, open-world segmentation , author=. Proceedings of the IEEE/CVF international conference on computer vision , pages=

  41. [48]

    IEEE transactions on image processing , volume=

    Image quality assessment: from error visibility to structural similarity , author=. IEEE transactions on image processing , volume=. 2004 , publisher=

    2004

  42. [49]

    Proceedings of the IEEE/CVF Winter conference on applications of computer vision , pages=

    Semantic-guided zero-shot learning for low-light image/video enhancement , author=. Proceedings of the IEEE/CVF Winter conference on applications of computer vision , pages=

  43. [50]

    2025 33rd European Signal Processing Conference (EUSIPCO) , pages=

    Zero-TIG: Temporal consistency-aware zero-shot illumination-guided low-light video enhancement , author=. 2025 33rd European Signal Processing Conference (EUSIPCO) , pages=. 2025 , organization=

    2025

  44. [51]

    Proceedings of the IEEE/CVF International Conference on Computer Vision , pages=

    Coherent event guided low-light video enhancement , author=. Proceedings of the IEEE/CVF International Conference on Computer Vision , pages=

  45. [52]

    Proceedings of the IEEE/CVF International Conference on Computer Vision , pages=

    NIR-assisted video enhancement via unpaired 24-hour data , author=. Proceedings of the IEEE/CVF International Conference on Computer Vision , pages=

  46. [53]

    Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) Workshops , month =

    Hu, Yuhuang and Liu, Shih-Chii and Delbruck, Tobi , title =. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) Workshops , month =. 2021 , pages =

    2021

  47. [55]

    IEEE Transactions on Multimedia , volume=

    Event-based low-illumination image enhancement , author=. IEEE Transactions on Multimedia , volume=. 2023 , publisher=

    2023

  48. [56]

    Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages=

    Towards robust event-guided low-light image enhancement: a large-scale real-world event-image dataset and novel approach , author=. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages=

  49. [57]

    Proceedings of the IEEE/CVF international conference on computer vision , pages=

    Foundir: Unleashing million-scale training data to advance foundation models for image restoration , author=. Proceedings of the IEEE/CVF international conference on computer vision , pages=

  50. [58]

    Proceedings of the Computer Vision and Pattern Recognition Conference , pages=

    Darkir: Robust low-light image restoration , author=. Proceedings of the Computer Vision and Pattern Recognition Conference , pages=

  51. [59]

    Cognitive Computation , volume=

    A joint network for low-light image enhancement based on retinex , author=. Cognitive Computation , volume=. 2024 , publisher=

    2024

  52. [60]

    International Journal of Computer Vision , volume=

    Modality-missing RGBT tracking: Invertible prompt learning and high-quality benchmarks , author=. International Journal of Computer Vision , volume=. 2025 , publisher=

    2025

  53. [61]

    Proceedings of the IEEE/CVF international conference on computer vision , pages=

    Robust monocular depth estimation under challenging conditions , author=. Proceedings of the IEEE/CVF international conference on computer vision , pages=

  54. [63]

    European Conference on Computer Vision , pages=

    Event enhanced high-quality image recovery , author=. European Conference on Computer Vision , pages=. 2020 , organization=

    2020

  55. [64]

    Proceedings of the AAAI Conference on Artificial Intelligence , volume=

    Low-light video enhancement with synthetic event guidance , author=. Proceedings of the AAAI Conference on Artificial Intelligence , volume=

  56. [65]

    Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages=

    Bracketing image restoration and enhancement with high-low frequency decomposition , author=. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages=

  57. [66]

    2025 IEEE 22nd International Symposium on Biomedical Imaging (ISBI) , pages=

    AMM-Diff: Adaptive Multi-Modality Diffusion Network for Missing Modality Imputation , author=. 2025 IEEE 22nd International Symposium on Biomedical Imaging (ISBI) , pages=. 2025 , organization=

    2025

  58. [67]

    Langley , title =

    P. Langley , title =. Proceedings of the 17th International Conference on Machine Learning (ICML 2000) , address =. 2000 , pages =

    2000

  59. [68]

    T. M. Mitchell. The Need for Biases in Learning Generalizations. 1980

    1980

  60. [69]

    M. J. Kearns , title =

  61. [70]

    Machine Learning: An Artificial Intelligence Approach, Vol. I. 1983

    1983

  62. [71]

    R. O. Duda and P. E. Hart and D. G. Stork. Pattern Classification. 2000

    2000

  63. [72]

    Suppressed for Anonymity , author=

  64. [73]

    Newell and P

    A. Newell and P. S. Rosenbloom. Mechanisms of Skill Acquisition and the Law of Practice. Cognitive Skills and Their Acquisition. 1981

    1981

  65. [74]

    A. L. Samuel. Some Studies in Machine Learning Using the Game of Checkers. IBM Journal of Research and Development. 1959

    1959

  66. [75]

    Learnable irrelevant modality dropout for multimodal action recognition on modality-specific annotated videos

    Alfasly, S., Lu, J., Xu, C., and Zou, Y. Learnable irrelevant modality dropout for multimodal action recognition on modality-specific annotated videos. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pp.\ 20208--20217, 2022

    2022

  67. [76]

    Retinexformer: One-stage retinex-based transformer for low-light image enhancement

    Cai, Y., Bian, H., Lin, J., Wang, H., Timofte, R., and Zhang, Y. Retinexformer: One-stage retinex-based transformer for low-light image enhancement. In Proceedings of the IEEE/CVF international conference on computer vision, pp.\ 12504--12513, 2023

    2023

  68. [77]

    N., and Koltun, V

    Chen, C., Chen, Q., Do, M. N., and Koltun, V. Seeing motion in the dark. In Proceedings of the IEEE/CVF International conference on computer vision, pp.\ 3185--3194, 2019

    2019

  69. [78]

    Bracketing image restoration and enhancement with high-low frequency decomposition

    Chen, G., Dai, K., Yang, K., Hu, T., Chen, X., Yang, Y., Dong, W., Wu, P., Zhang, Y., and Yan, Q. Bracketing image restoration and enhancement with high-low frequency decomposition. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp.\ 6097--6107, 2024 a

    2024

  70. [79]

    Evlight++: Low-light video enhancement with an event camera: A large-scale real-world dataset, novel method, and more

    Chen, K., Liang, G., Lu, Y., Li, H., and Wang, L. Evlight++: Low-light video enhancement with an event camera: A large-scale real-world dataset, novel method, and more. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2025 a

    2025

  71. [80]

    Frequency-aware feature fusion for dense image prediction

    Chen, L., Fu, Y., Gu, L., Yan, C., Harada, T., and Huang, G. Frequency-aware feature fusion for dense image prediction. IEEE transactions on pattern analysis and machine intelligence, 2024 b

    2024

  72. [81]

    Frequency dynamic convolution for dense image prediction

    Chen, L., Gu, L., Li, L., Yan, C., and Fu, Y. Frequency dynamic convolution for dense image prediction. In Proceedings of the Computer Vision and Pattern Recognition Conference, pp.\ 30178--30188, 2025 b

    2025

  73. [82]

    Low light video enhancement by learning on static videos with cross-frame attention

    Chhirolya, S., Malik, S., and Soundararajan, R. Low light video enhancement by learning on static videos with cross-frame attention. arXiv preprint arXiv:2210.04290, 2022

    work page arXiv 2022

  74. [83]

    Unbiased missing-modality multimodal learning

    Dai, R., Li, C., Yan, Y., Mo, L., Qin, K., and He, T. Unbiased missing-modality multimodal learning. In Proceedings of the IEEE/CVF International Conference on Computer Vision, pp.\ 24507--24517, 2025

    2025

  75. [84]

    arXiv preprint arXiv:2508.05630 (2025) 8

    Ding, H., Ying, K., Liu, C., He, S., Jiang, X., Jiang, Y.-G., Torr, P. H., and Bai, S. Mosev2: A more challenging dataset for video object segmentation in complex scenes. arXiv preprint arXiv:2508.05630, 2025

    work page arXiv 2025

  76. [85]

    C., Garcia, A., and Conde, M

    Feijoo, D., Benito, J. C., Garcia, A., and Conde, M. V. Darkir: Robust low-light image restoration. In Proceedings of the Computer Vision and Pattern Recognition Conference, pp.\ 10879--10889, 2025

    2025

  77. [86]

    Dancing in the dark: A benchmark towards general low-light video enhancement

    Fu, H., Zheng, W., Wang, X., Wang, J., Zhang, H., and Ma, H. Dancing in the dark: A benchmark towards general low-light video enhancement. In Proceedings of the IEEE/CVF International Conference on Computer Vision, pp.\ 12877--12886, 2023

    2023

  78. [87]

    Robust monocular depth estimation under challenging conditions

    Gasperini, S., Morbitzer, N., Jung, H., Navab, N., and Tombari, F. Robust monocular depth estimation under challenging conditions. In Proceedings of the IEEE/CVF international conference on computer vision, pp.\ 8177--8186, 2023

    2023

  79. [88]

    Metabev: Solving sensor failures for 3d detection and map segmentation

    Ge, C., Chen, J., Xie, E., Wang, Z., Hong, L., Lu, H., Li, Z., and Luo, P. Metabev: Solving sensor failures for 3d detection and map segmentation. In Proceedings of the IEEE/CVF International Conference on Computer Vision, pp.\ 8721--8731, 2023

    2023

  80. [89]

    v2e: From video frames to realistic dvs events

    Hu, Y., Liu, S.-C., and Delbruck, T. v2e: From video frames to realistic dvs events. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) Workshops, pp.\ 1312--1321, June 2021

    2021

Showing first 80 references.