Pith. sign in

REVIEW 4 major objections 5 minor 45 references

Implicit Neural Representation for Video Restoration

T0 review · 4 major / 5 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 A network trained only on clean ×4 video pairs claims to upscale at any scale and denoise zero-shot.

desk verdict Admitted per-video fine-tuning in the conclusion guts the zero-shot and arbitrary-scale claims; the 44 dB numbers are fits, not predictions. read the letter →

arxiv 2506.05488 v1 pith:E5X7YNJU submitted 2025-06-05 cs.CV

Mary Aiyetigbo , Wanqi Yuan , Feng Luo , Nianyi Li This is my paper
classification cs.CV
keywords implicitneuralrepresentationvideosuper-resolutionzero-shotdenoisingarbitrary-scalerestorationmulti-resolutionhashencodingcoordinate-basednetworkspixel-erroramplifiedloss
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

VR-INR claims that one implicit neural network, trained only on clean low- and high-resolution video pairs at a single upscaling factor (×4), can at test time produce high-resolution frames at arbitrary unseen scales (from ×2 to ×32) and suppress Gaussian and Poisson noise it has never seen. The method replaces explicit optical flow with a hierarchical spatial–temporal–texture encoding and multi-resolution hash interpolation, decoding each output pixel from a continuous coordinate representation. If the claim holds, video restoration no longer needs a separate model per scale or per degradation, and clean data alone may suffice to handle noisy inputs. The paper reports consistently higher PSNR/SSIM than fixed-scale baselines across Vid4, REDS4, GOPRO, and DAVIS at ×2, ×4, and ×8.

What carries the argument

The load-bearing object is a multi-resolution implicit hash encoding over a 6-dimensional coordinate-texture space. For each target high-resolution coordinate, the network retrieves 64 neighboring hash-table features, combines them with learned adaptive weights, refines features from coarse to fine with a top-down attention mechanism, and decodes the result to RGB. Because the map from coordinates to RGB is continuous, the same weights can be queried at any upscaling factor; a pixel-error amplified loss additionally up-weights low-error pixels so fine detail is not smoothed away by MSE.

What would settle it

Run the published evaluation protocol exactly as described in Section 4.2 but disable any per-video fine-tuning before testing; if the ×2 and ×8 PSNR values on Vid4 exceed the baselines by the reported margins, the zero-shot claim stands, and if they fall to baseline levels, the claim is falsified.

Watch

Extended reading notes

Core claim

The paper's central claim is that VR-INR is a single network that performs both arbitrary-scale video super-resolution and zero-shot denoising after being trained only on clean ×4 LR-HR pairs. The authors argue that representing video as a continuous function over a 6-dimensional spatial-temporal-texture space, with features read from a multi-resolution hash table and decoded by an MLP, is what lets the same model answer queries at ×2 through ×32 and remove unseen noise without retraining or motion estimation. They support this with quantitative comparisons showing large PSNR/SSIM advantages over VideoINR, VRT, IART, and SAVSR at in-distribution ×4 and out-of-distribution ×2/×8, and with zero-shot denoising results on DAVIS (e.g., around 31.5 dB at σ=30 versus roughly 19.9 dB for the strongest baseline).

Load-bearing premise

The load-bearing premise is that the reported generalization comes from the single ×4-trained network; the conclusion says VR-INR is fine-tuned per video with clean LR-HR pairs, and if that per-video adaptation is required to reach the reported PSNR values, the zero-shot and arbitrary-scale claims would instead describe test-time fitting.

Editorial extensions

If this is right

  • Video super-resolution systems could serve arbitrary user-selected zoom factors from one trained model, eliminating per-scale retraining and model zoo maintenance.
  • Denoising could be obtained as a byproduct of clean-data super-resolution training, reducing the need for paired noisy and clean video data.
  • The flow-free design avoids optical-flow failures on occlusions and large motion, simplifying the restoration pipeline.
  • The pixel-error amplified loss is a transferable training objective for any coordinate-based reconstruction task where standard MSE over-smooths edges and textures.

Reading between the lines

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

  • The conclusion mentions per-video fine-tuning with clean LR-HR pairs, so the public numbers should be reported both with and without that adaptation to separate learned generalization from test-time fitting; this distinction is an open question the paper leaves implicit.
  • If the scale generalization is genuine, it likely comes from the multiresolution hash grids acting as a scale-aware dictionary; training the same architecture at multiple scales would be a natural stress test, since multi-scale training should improve the arbitrary-scale curve if the single-scale prior is already saturated.
  • Zero-shot denoising may be an emergent bias of the coordinate decoder toward smooth, low-frequency outputs; measuring output high-frequency energy on noisy inputs, or adding a small noise-removal head, would clarify the mechanism.
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

4 major / 5 minor

Summary. The paper proposes VR-INR, an implicit neural representation framework for video restoration, claiming that a single network trained only on clean LR-HR pairs at a fixed ×4 upscaling factor generalizes to arbitrary unseen super-resolution scales and performs zero-shot denoising. The method combines hierarchical spatial-temporal-texture encoding, multi-resolution hash embedding, a top-down attention mechanism, and a proposed pixel-error amplified loss. Experiments on Vid4, REDS4, GOPRO, and DAVIS report large improvements over prior video super-resolution and denoising baselines. However, the conclusion states that VR-INR is fine-tuned per video using clean LR-HR pairs, which directly contradicts the abstract and inference sections' claims of no additional training and zero-shot generalization.

Significance. If the central claim were valid—a single model trained only on clean ×4 pairs that generalizes to arbitrary scales and unseen noise—the contribution would be significant for video restoration. The proposed architecture and loss may contain useful ideas. However, the admitted per-video fine-tuning in Sec. 5 undermines the central generalization claim: the reported evaluations may measure per-sequence fitting rather than the behavior of a once-trained model. The reported PSNR values (e.g., 44.21 dB on Vid4 ×4, 46.93 dB in the ablation) are implausibly high relative to published SOTA, and the experimental section does not describe a protocol that would make them reproducible. The paper does not ship code, checkpoints, or detailed experimental settings, and the internal inconsistency between the abstract and conclusion is load-bearing rather than cosmetic.

major comments (4)
  1. [Section 5 (Conclusion)] The conclusion states: 'By fine-tuning per video using only clean LR–HR pairs, VR-INR adapts to each sequence's unique content and noise characteristics.' This directly contradicts the abstract and Sec. 3.5, which claim operation 'without additional training' and 'zero-shot denoising.' If the model is fine-tuned on each test video, the reported results are fitted values for each sequence, not generalization of a model trained once on clean ×4 pairs. The paper must specify precisely which parameters are trained once and which are optimized per video, and must re-run the evaluation without any test-time adaptation to support the stated central claim.
  2. [Section 4.2, Table 1] The reported PSNR values are far outside the range of published video super-resolution results. For example, the paper reports 44.21 dB on Vid4 ×4 and 42.00 dB on DAVIS ×4, whereas typical SOTA on Vid4 ×4 is around 28 dB and SA-VSR (a strong arbitrary-scale method) is reported at 30.95 dB ×2 and 24.50 dB ×4 in the paper's own table. No error bars, number of test clips, or train/test separation are provided. The experimental section does not describe whether the reported numbers come from the once-trained model or from per-video fine-tuning. Without this information, the comparison with baselines is uninterpretable.
  3. [Section 4.3, Table 4] The ablation study reports 46.93 dB PSNR on DAVIS ×4, which is 4.93 dB higher than the 42.00 dB reported for the same dataset and scale in Table 1. If the higher number arises from per-video fine-tuning in the ablation while Table 1 uses a single fixed model, the paper must clarify this; if not, the inconsistency suggests that the reported metrics are not reproducible from the described experimental settings. The paper should report the evaluation protocol for each table explicitly.
  4. [Section 4.2, 'Compared with SOTAs'] The baselines (VideoINR, VRT, IART, SA-VSR) are evaluated without any fine-tuning at unseen scales or noise levels, while VR-INR is fine-tuned per video according to Sec. 5. This creates an unfair comparison: the numbers attributed to VR-INR may reflect test-time adaptation that the baselines are not allowed. The paper must either evaluate all methods under the same protocol (i.e., no per-video fine-tuning) or separate results into 'zero-shot' and 'fine-tuned' conditions.
minor comments (5)
  1. [Section 3.5, Eq. (17)] Equation (17) contains malformed notation with nested and garbled parentheses that make the expression unreadable; it should be rewritten with clear definitions of all subterms.
  2. [Throughout] The dataset name 'DA VIS' should be 'DAVIS' (as in the DAVIS challenge reference), and the method name 'NERV' in Table 3 and related text should be 'NeRV'.
  3. [Section 4.2, Dataset paragraph] The sentence 'For DA VIS and GOPRO, we first resized the video frames to 256 × 256 pixels, which served as the HR ground truth' is unclear about the generation of LR inputs; specify the exact downsampling procedure used to produce LR frames from the HR ground truth.
  4. [Section 4.3, PEA-loss hyperparameters] The text claims that the PEA-loss hyperparameters τ, ε, and δ have little impact within small ranges, but no quantitative ablation table is provided for these values. The selected values (τ=0.01, ε=0.005, δ=0.001, α=5) should be supported by a table or at least a figure.
  5. [Section 5, Conclusion] The conclusion claims the method is 'computationally efficient' and 'simplifies the restoration pipeline,' but no runtime, parameter count, or FLOPs comparisons are given anywhere in the paper.

Circularity Check

1 steps flagged · score 8.0 of 10

Admitted per-video fine-tuning in Sec. 5 turns the 'zero-shot' and 'unseen scale' results into per-sequence fitted values, so the central predictions reduce by construction.

  1. fitted input called prediction [Section 5 (Conclusion), contradicted by Abstract and Section 3.5]
    "By fine-tuning per video using only clean LR–HR pairs, VR-INR adapts to each sequence's unique content and noise characteristics, delivering superior PSNR and SSIM on Vid4, REDS4, GOPRO, and DA VIS—even under unseen noise levels."

    The abstract claims VR-INR 'is trained only on a single upscaling factor (×4)' and performs 'zero-shot denoising on noisy input, despite never having seen noisy data during training'; Sec. 3.5 adds that it does so 'without additional training.' The conclusion, however, states the actual evaluation protocol is per-video fine-tuning with clean LR-HR pairs of each test sequence. Fine-tuning on the same video's clean HR frames (and, for denoising, on the same video's noisy LR paired with clean HR) means every reported PSNR/SSIM is produced by a model adapted to that exact sequence, not by the once-trained ×4 network.

full rationale

The central claim is that a single model trained only on clean ×4 LR-HR pairs generalizes to arbitrary scales and denoises zero-shot. The paper's own conclusion discloses that the model is fine-tuned per test video using clean LR-HR pairs, which is supervised test-time adaptation on the evaluation sequence. This directly contradicts the abstract and Sec. 3.5 claims of 'without additional training' and 'never having seen noisy data.' The quantitative results (e.g., 44.21 dB on Vid4 ×4, 31.50 dB zero-shot denoising) are implausible for a once-trained model and are consistent with per-sequence fitting. Related Work cites the authors' own prior per-video INR fitting work [1] as the denoising mechanism, and the conclusion confirms that mechanism is load-bearing; the 'zero-shot' novelty label is thus a renaming of per-video fitting. No other circularity was found in the architecture or loss derivation, but the core empirical claim reduces to the fitted values by the paper's own protocol, so the circularity score is high.

Assumptions & free parameters 5 free parameters · 3 assumptions · 0 invented entities

The method introduces no new physical or mathematical entities, but the protocol depends on an implicit per-video fitting (fine-tuning) that is not stated as a constraint in the abstract. The free parameters are mostly loss hyperparameters, and the key unknown is the feature code length inconsistency.

free parameters (5)
  • reconstruction mask threshold (tau) = 0.01
    Hyperparameter in PEA-loss (Eq. 13); tuned on DAVIS, affects which pixels are masked.
  • boosting error threshold (epsilon) = 0.005
    Hyperparameter in PEA-loss (Eq. 15); tuned on DAVIS.
  • boosting constant (delta) = 0.001
    Hyperparameter in PEA-loss (Eq. 15); tuned on DAVIS.
  • boosting weight factor (alpha) = 5
    Hyperparameter in PEA-loss (Eq. 16); authors state it has the most significant impact and was selected by extensive experimentation.
  • feature code length (F) = 3 in method, 6 in ablation
    The paper states F=3 in Sec. 3.1, but the ablation Table 4 varies F from 4 to 7 and reports best at 6. The inconsistency makes the actual configuration unclear.
assumptions (3)
  • domain assumption Bicubic upsampling of the LR input provides a suitable initialization from which the hash table and MLPs can learn to reconstruct HR details.
    Used in Sec. 3.1 and at inference: 'we first resize the LR frames to the target resolution using bicubic interpolation' (Eq. 1). The method assumes this upsampled image contains enough information for the coordinate network to refine.
  • ad hoc to paper A per-video hash table can memorize a video's content and that this memorization is what the reported PSNR measures.
    The conclusion states 'fine-tuning per video using only clean LR-HR pairs'. This is an unstated assumption in the experimental protocol; without it, the ablation numbers (e.g. PSNR 46.93) would be impossible.
  • domain assumption The 6-dimensional STT code, including 3 texture features, is a sufficient representation for all video content.
    Sec. 3.1 defines the STT code as [x, y, t, xi1, xi2, xi3] with F=3. The entire method rests on the sufficiency of this compact code for reconstructing high-frequency details.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Implicit Neural Representation for Video Restoration." pith.science (2026). https://pith.science/paper/E5X7YNJU

@misc{pith2026250605488,
  author       = {Pith},
  title        = {Pith review of: Implicit Neural Representation for Video Restoration},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/E5X7YNJU}},
  note         = {Machine review of arXiv:2506.05488}
}
abstract

High-resolution (HR) videos play a crucial role in many computer vision applications. Although existing video restoration (VR) methods can significantly enhance video quality by exploiting temporal information across video frames, they are typically trained for fixed upscaling factors and lack the flexibility to handle scales or degradations beyond their training distribution. In this paper, we introduce VR-INR, a novel video restoration approach based on Implicit Neural Representations (INRs) that is trained only on a single upscaling factor ($\times 4$) but generalizes effectively to arbitrary, unseen super-resolution scales at test time. Notably, VR-INR also performs zero-shot denoising on noisy input, despite never having seen noisy data during training. Our method employs a hierarchical spatial-temporal-texture encoding framework coupled with multi-resolution implicit hash encoding, enabling adaptive decoding of high-resolution and noise-suppressed frames from low-resolution inputs at any desired magnification. Experimental results show that VR-INR consistently maintains high-quality reconstructions at unseen scales and noise during training, significantly outperforming state-of-the-art approaches in sharpness, detail preservation, and denoising efficacy.

Figures

Figures reproduced from arXiv: 2506.05488 by the authors.

Figure 1
Figure 1. VR-INR demonstrates robust scale generalization and zero-shot denoising for video restoration. Although trained only on clean LR–HR pairs at ×4, VR-INR generalizes to arbitrary unseen scales (e.g., ×2, ×8, ×16) and removes noise from degraded inputs without any noise-specific training, producing high-quality restoration across scales and noise levels. Bottom row: VR-INR outputs; green labels denote out-of-distributi… view at source ↗
Figure 2
Figure 2. VR-INR training pipeline. Local patches are extracted at multiple resolutions and processed [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. The video super resolution effectiveness of [PITH_FULL_IMAGE:figures/full_fig_p007_3.png] view at source ↗
Figures from the paper (14 more)
Figure 4
Figure 4. Figure 4: Visual comparison with state-of-the-art methods on the Vid4 [ [PITH_FULL_IMAGE:figures/full_fig_p008_4.png]
Figure 5
Figure 5. Figure 5: Visual comparison with state-of-the-art (SOTA) video super-resolution (VSR) methods on [PITH_FULL_IMAGE:figures/full_fig_p008_5.png]
Figure 6
Figure 6. Figure 6: Results of NERV and Ours. From top to bottom: Vid4 and GOPRO datasets [PITH_FULL_IMAGE:figures/full_fig_p008_6.png]
Figure 7
Figure 7. Figure 7: Visual comparison of the impact of our top-down attention mechanism and the [PITH_FULL_IMAGE:figures/full_fig_p009_7.png]
Figure 8
Figure 8. Figure 8: Visual comparison of our model against state-of-the-art methods on the REDS4 dataset for [PITH_FULL_IMAGE:figures/full_fig_p013_8.png]
Figure 9
Figure 9. Figure 9: Visual comparison of our model against state-of-the-art methods on the GOPRO dataset for [PITH_FULL_IMAGE:figures/full_fig_p013_9.png]
Figure 10
Figure 10. Figure 10: Visual comparison of for x2 scale factor on the DAVIS dataset. [PITH_FULL_IMAGE:figures/full_fig_p014_10.png]
Figure 11
Figure 11. Figure 11: Visual comparison of our model against state-of-the-art methods across various arbitrary [PITH_FULL_IMAGE:figures/full_fig_p015_11.png]
Figure 12
Figure 12. Figure 12: Visual comparison to show the effectiveness of our model for performing zero-shot [PITH_FULL_IMAGE:figures/full_fig_p015_12.png]
Figure 13
Figure 13. Figure 13: Visual comparison to show the effectiveness of our model for performing zero-shot [PITH_FULL_IMAGE:figures/full_fig_p016_13.png]
Figure 14
Figure 14. Figure 14: Visual comparison of video reconstruction on VID4 video dataset. [PITH_FULL_IMAGE:figures/full_fig_p016_14.png]
Figure 15
Figure 15. Figure 15: Visual comparison of video reconstruction on GOPRO video dataset. [PITH_FULL_IMAGE:figures/full_fig_p017_15.png]
Figure 16
Figure 16. Figure 16: Visual comparison of zero-shot denoising results using our video reconstruction framework [PITH_FULL_IMAGE:figures/full_fig_p017_16.png]
Figure 17
Figure 17. Figure 17: Visual comparison of zero-shot denoising results using our video reconstruction framework [PITH_FULL_IMAGE:figures/full_fig_p018_17.png]

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

45 extracted references · 42 canonical work pages

  1. [1]

    Unsupervised coordinate-based video denoising

    Mary Aiyetigbo, Dineshchandar Ravichandran, Reda Chalhoub, Peter Kalivas, Feng Luo, and Nianyi Li. Unsupervised coordinate-based video denoising. In 2024 IEEE International Conference on Image Processing (ICIP), pages 1438–1444. IEEE, 2024

    work page 2024

  2. [2]

    Basicvsr: The search for essential components in video super-resolution and beyond

    Kelvin CK Chan, Xintao Wang, Ke Yu, Chao Dong, and Chen Change Loy. Basicvsr: The search for essential components in video super-resolution and beyond. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 4947–4956, 2021

    work page 2021

  3. [3]

    Basicvsr++: Improving video super-resolution with enhanced propagation and alignment

    Kelvin CK Chan, Shangchen Zhou, Xiangyu Xu, and Chen Change Loy. Basicvsr++: Improving video super-resolution with enhanced propagation and alignment. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 5972–5981, 2022

    work page 2022

  4. [4]

    Learning continuous image representation with local implicit image function

    Yinbo Chen, Sifei Liu, and Xiaolong Wang. Learning continuous image representation with local implicit image function. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 8628–8638, 2021

    work page 2021

  5. [5]

    Videoinr: Learning video implicit neural representation for continuous space-time super- resolution

    Zeyuan Chen, Yinbo Chen, Jingwen Liu, Xingqian Xu, Vidit Goel, Zhangyang Wang, Humphrey Shi, and Xiaolong Wang. Videoinr: Learning video implicit neural representation for continuous space-time super- resolution. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 2047–2057, 2022

    work page 2022

  6. [6]

    Nerv: Neural representations for videos

    Zhengyu Chen, Hexiang Wu, and Yuan-Fang Wang. Nerv: Neural representations for videos. In Advances in Neural Information Processing Systems, 2021

    work page 2021

  7. [7]

    Learning spatial adaptation and temporal coherence in diffusion models for video super-resolution

    Zhikai Chen, Fuchen Long, Zhaofan Qiu, Ting Yao, Wengang Zhou, Jiebo Luo, and Tao Mei. Learning spatial adaptation and temporal coherence in diffusion models for video super-resolution. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 9232–9241, 2024

    work page 2024

  8. [8]

    Distilling information with super-resolution for video surveillance

    Marco Cristani, Dong Seon Cheng, Vittorio Murino, and Donato Pannullo. Distilling information with super-resolution for video surveillance. In Proceedings of the ACM 2nd international workshop on Video surveillance & sensor networks, pages 2–11, 2004

    work page 2004

Show all 45 references
  1. [9]

    Super-resolution in medical imaging

    Hayit Greenspan. Super-resolution in medical imaging. The computer journal, 52(1):43–63, 2009

    2009

  2. [10]

    Hnerv: A hybrid neural representation for videos

    Rui Han, Zhengyu Chen, and Yuan-Fang Wang. Hnerv: A hybrid neural representation for videos. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 18966– 18975, 2022

    2022

  3. [11]

    Recurrent back-projection network for video super-resolution

    Muhammad Haris, Gregory Shakhnarovich, and Norimichi Ukita. Recurrent back-projection network for video super-resolution. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 3897–3906, 2019

    2019

  4. [12]

    Video super-resolution via bidirectional recurrent convolutional networks

    Yan Huang, Wei Wang, and Liang Wang. Video super-resolution via bidirectional recurrent convolutional networks. IEEE transactions on pattern analysis and machine intelligence, 40(4):1015–1028, 2017

    2017

  5. [13]

    Tempdiff: Enhancing temporal-awareness in latent diffusion for real-world video super-resolution

    Qin Jiang, Qing Lin Wang, Li Hua Chi, Xin Hai Chen, Qing Yang Zhang, Richard Zhou, Zheng Qiu Deng, Jin Sheng Deng, Bin Bing Tang, Shao He Lv, et al. Tempdiff: Enhancing temporal-awareness in latent diffusion for real-world video super-resolution. Computer Graphics Forum, 43(7)...

    2024

  6. [14]

    Deep video super-resolution network using dynamic upsampling filters without explicit motion compensation

    Younghyun Jo, Seoung Wug Oh, Jaeyeon Kang, and Seon Joo Kim. Deep video super-resolution network using dynamic upsampling filters without explicit motion compensation. In Proceedings of the IEEE conference on computer vision and pattern recognition, pages 3224–3232, 2018

    2018

  7. [15]

    Video super-resolution with convolutional neural networks

    Armin Kappeler, Seunghwan Yoo, Qiqin Dai, and Aggelos K Katsaggelos. Video super-resolution with convolutional neural networks. IEEE transactions on computational imaging, 2(2):109–122, 2016

    2016

  8. [16]

    A review of the deep learning methods for medical images super resolution problems

    Yufei Li, Bruno Sixou, and Francois Peyrin. A review of the deep learning methods for medical images super resolution problems. Irbm, 42(2):120–133, 2021

    2021

  9. [17]

    Savsr: arbitrary-scale video super-resolution via a learned scale-adaptive network

    Zekun Li, Hongying Liu, Fanhua Shang, Yuanyuan Liu, Liang Wan, and Wei Feng. Savsr: arbitrary-scale video super-resolution via a learned scale-adaptive network. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 38, pages 3288–3296, 2024

    2024

  10. [18]

    Vrt: A video restoration transformer

    Jingyun Liang, Jiezhang Cao, Yuchen Fan, Kai Zhang, Rakesh Ranjan, Yawei Li, Radu Timofte, and Luc Van Gool. Vrt: A video restoration transformer. IEEE Transactions on Image Processing, 2024. 10

    2024

  11. [19]

    Recurrent video restoration transformer with guided deformable attention

    Jingyun Liang, Yuchen Fan, Xiaoyu Xiang, Rakesh Ranjan, Eddy Ilg, Simon Green, Jiezhang Cao, Kai Zhang, Radu Timofte, and Luc V Gool. Recurrent video restoration transformer with guided deformable attention. Advances in Neural Information Processing Systems, 35:378–393, 2022

    2022

  12. [20]

    A bayesian approach to adaptive video super resolution

    Ce Liu and Deqing Sun. A bayesian approach to adaptive video super resolution. In The 24th IEEE Conference on Computer Vision and Pattern Recognition, CVPR 2011, Colorado Springs, CO, USA, 20-25 June 2011, pages 209–216. IEEE Computer Society, 2011

    2011

  13. [21]

    Vsrdiff: Learning inter-frame temporal coherence in diffusion model for video super-resolution

    Linlin Liu, Lele Niu, Jun Tang, and Yong Ding. Vsrdiff: Learning inter-frame temporal coherence in diffusion model for video super-resolution. IEEE Access, 2025

    2025

  14. [22]

    Stadnet: Spatial-temporal attention-guided dual-path network for cardiac cine mri super- resolution

    Jun Lyu, Shuo Wang, Yapeng Tian, Jing Zou, Shunjie Dong, Chengyan Wang, Angelica I Aviles-Rivero, and Jing Qin. Stadnet: Spatial-temporal attention-guided dual-path network for cardiac cine mri super- resolution. Medical Image Analysis, 94:103142, 2024

    2024

  15. [23]

    Instant neural graphics primitives with a multiresolution hash encoding

    Thomas Müller, Alex Evans, Christoph Schied, and Alexander Keller. Instant neural graphics primitives with a multiresolution hash encoding. ACM Transactions on Graphics, 41(4):102:1–102:15, 2022

    2022

  16. [24]

    Ntire 2019 challenge on video deblurring and super-resolution: Dataset and study

    Seungjun Nah, Sungyong Baik, Seokil Hong, Gyeongsik Moon, Sanghyun Son, Radu Timofte, and Kyoung Mu Lee. Ntire 2019 challenge on video deblurring and super-resolution: Dataset and study. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition works...

    2019

  17. [25]

    Deep multi-scale convolutional neural network for dynamic scene deblurring

    Seungjun Nah, Tae Hyun Kim, and Kyoung Mu Lee. Deep multi-scale convolutional neural network for dynamic scene deblurring. In 2017 IEEE Conference on Computer Vision and Pattern Recognition, CVPR 2017, Honolulu, HI, USA, July 21-26, 2017, pages 257–265. IEEE Computer Society, 2017

    2017

  18. [26]

    Super-resolution image reconstruction: a technical overview

    Sung Cheol Park, Min Kyu Park, and Moon Gi Kang. Super-resolution image reconstruction: a technical overview. IEEE Signal Processing Magazine, 20(3):21–36, 2003

    2003

  19. [27]

    The 2017 davis challenge on video object segmentation

    Jordi Pont-Tuset, Federico Perazzi, Sergi Caelles, Pablo Arbeláez, Alex Sorkine-Hornung, and Luc Van Gool. The 2017 davis challenge on video object segmentation. arXiv preprint arXiv:1704.00675, 2017

    2017 arXiv

  20. [28]

    Frame-recurrent video super-resolution

    Mehdi SM Sajjadi, Raviteja Vemulapalli, and Matthew Brown. Frame-recurrent video super-resolution. In Proceedings of the IEEE conference on computer vision and pattern recognition, pages 6626–6634, 2018

    2018

  21. [29]

    Rethinking alignment in video super-resolution transformers

    Shuwei Shi, Jinjin Gu, Liangbin Xie, Xintao Wang, Yujiu Yang, and Chao Dong. Rethinking alignment in video super-resolution transformers. Advances in Neural Information Processing Systems, 35:36081–36093, 2022

    2022

  22. [30]

    Con- volutional lstm network: A machine learning approach for precipitation nowcasting

    Xingjian Shi, Zhourong Chen, Hao Wang, Dit-Yan Yeung, Wai-Kin Wong, and Wang-chun Woo. Con- volutional lstm network: A machine learning approach for precipitation nowcasting. Advances in neural information processing systems, 28, 2015

    2015

  23. [31]

    Martel, Alexander W

    Vincent Sitzmann, Julien N.P. Martel, Alexander W. Bergman, David B. Lindell, and Gordon Wetzstein. Implicit neural representations with periodic activation functions. In Advances in Neural Information Processing Systems, volume 33, pages 7462–7473, 2020

    2020

  24. [32]

    Srinivasan, Ben Mildenhall, Sara Fridovich-Keil, Neel Raghavan, Utkarsh Singhal, Ravi Ramamoorthi, Jonathan T

    Matthew Tancik, Pratul P. Srinivasan, Ben Mildenhall, Sara Fridovich-Keil, Neel Raghavan, Utkarsh Singhal, Ravi Ramamoorthi, Jonathan T. Barron, and Ren Ng. Fourier features let networks learn high frequency functions in low dimensional domains. In Advances in Neural Informati...

    2020

  25. [33]

    Tdan: Temporally-deformable alignment network for video super-resolution

    Yapeng Tian, Yulun Zhang, Yun Fu, and Chenliang Xu. Tdan: Temporally-deformable alignment network for video super-resolution. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 3360–3369, 2020

    2020

  26. [34]

    Deep image prior

    Dmitry Ulyanov, Andrea Vedaldi, and Victor Lempitsky. Deep image prior. InProceedings of the IEEE conference on computer vision and pattern recognition, pages 9446–9454, 2018

    2018

  27. [35]

    Edvr: Video restoration with enhanced deformable convolutional networks

    Xintao Wang, Kelvin CK Chan, Ke Yu, Chao Dong, and Chen Change Loy. Edvr: Video restoration with enhanced deformable convolutional networks. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition workshops, pages 0–0, 2019

    2019

  28. [36]

    Video dehazing via a multi-range temporal alignment network with physical prior

    Jiaqi Xu, Xiaowei Hu, Lei Zhu, Qi Dou, Jifeng Dai, Yu Qiao, and Pheng-Ann Heng. Video dehazing via a multi-range temporal alignment network with physical prior. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 18053–18062, 2023. 11

    2023

  29. [37]

    Enhancing video super-resolution via implicit resampling-based alignment

    Kai Xu, Ziwei Yu, Xin Wang, Michael Bi Mi, and Angela Yao. Enhancing video super-resolution via implicit resampling-based alignment. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 2546–2555, 2024

    2024

  30. [38]

    Motion-guided latent diffusion for temporally consistent real-world video super-resolution

    Xi Yang, Chenhang He, Jianqi Ma, and Lei Zhang. Motion-guided latent diffusion for temporally consistent real-world video super-resolution. In European Conference on Computer Vision, pages 224–242. Springer, 2025

    2025

  31. [39]

    Omniscient video super-resolution

    Peng Yi, Zhongyuan Wang, Kui Jiang, Junjun Jiang, Tao Lu, Xin Tian, and Jiayi Ma. Omniscient video super-resolution. In Proceedings of the IEEE/CVF international conference on computer vision, pages 4429–4438, 2021

    2021

  32. [40]

    Inflation with diffusion: Efficient temporal adaptation for text-to-video super-resolution

    Xin Yuan, Jinoo Baek, Keyang Xu, Omer Tov, and Hongliang Fei. Inflation with diffusion: Efficient temporal adaptation for text-to-video super-resolution. In Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision, pages 489–496, 2024

    2024

  33. [41]

    Super-resolution reconstruction for multi-angle remote sensing images considering resolution differences

    Hongyan Zhang, Zeyu Yang, Liangpei Zhang, and Huanfeng Shen. Super-resolution reconstruction for multi-angle remote sensing images considering resolution differences. Remote Sensing, 6(1):637–657, 2014

    2014

  34. [42]

    Feature consistency-based style transfer for landscape images using dual-channel attention

    Qiang Zhang, Shuai Wang, and Dong Cui. Feature consistency-based style transfer for landscape images using dual-channel attention. IEEE Access, 2024

    2024

  35. [43]

    Avernet: All-in-one video restoration for time-varying unknown degradations

    Haiyu Zhao, Lei Tian, Xinyan Xiao, Peng Hu, Yuanbiao Gou, and Xi Peng. Avernet: All-in-one video restoration for time-varying unknown degradations. Advances in Neural Information Processing Systems, 37:127296–127316, 2024

    2024

  36. [44]

    Upscale-a-video: Temporal-consistent diffusion model for real-world video super-resolution

    Shangchen Zhou, Peiqing Yang, Jianyi Wang, Yihang Luo, and Chen Change Loy. Upscale-a-video: Temporal-consistent diffusion model for real-world video super-resolution. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 2535–2545, 2024. 1...

    2024

  37. [45]

    These comparisons span video super-resolution tasks at scale factors of ×2, ×4, and ×8, producing outputs at a resolution of 256×256

    datasets. These comparisons span video super-resolution tasks at scale factors of ×2, ×4, and ×8, producing outputs at a resolution of 256×256. As illustrated in Figure 8, our model demonstrates superior preservation of fine textures and structural details on the REDS4 dataset...

Pith tools

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