pith. sign in

arxiv: 2502.06431 · v2 · submitted 2025-02-10 · 💻 cs.CV

FCVSR: A Frequency-aware Method for Compressed Video Super-Resolution

Qiang Zhu , Fan Zhang , Feiyu Chen , Shuyuan Zhu , David Bull , Bing Zeng This is my paper

Pith reviewed 2026-05-23 03:46 UTC · model grok-4.3

classification 💻 cs.CV
keywords compressed video super-resolutionfrequency domainmotion alignmentfeature refinementcontrastive lossdeep neural networksspatial frequency subbandstemporal dynamics
0
0 comments X

The pith

FCVSR processes frequency subbands separately in space and tracks their temporal changes to improve compressed video super-resolution.

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

The paper introduces FCVSR to generate high-resolution videos from low-resolution compressed inputs by addressing gaps in existing frequency-domain approaches. Prior methods overlook spatial differences across frequency subbands and fail to track how those frequencies evolve over time, which the authors argue produces suboptimal outputs. FCVSR adds a motion-guided adaptive alignment network to handle motion, a multi-frequency feature refinement module to process subbands distinctly, and a frequency-aware contrastive loss to sharpen details. Tests on three public datasets report up to 0.14 dB higher PSNR than the next best method at lower complexity.

Core claim

FCVSR consists of a motion-guided adaptive alignment network and a multi-frequency feature refinement module, trained with a frequency-aware contrastive loss, to differentiate frequency subbands spatially while capturing temporal frequency dynamics for compressed video super-resolution.

What carries the argument

The multi-frequency feature refinement module, which separates and refines distinct frequency subbands while incorporating motion alignment and contrastive training.

If this is right

  • The model delivers up to 0.14 dB PSNR improvement over the second-best method on standard compressed video SR benchmarks.
  • Model complexity stays lower than competing approaches while achieving the reported quality gains.
  • The frequency-aware contrastive loss enables reconstruction of finer spatial details than standard losses.
  • Spatio-temporal frequency information is handled more effectively than in earlier frequency-domain video SR techniques.

Where Pith is reading between the lines

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

  • The same subband separation and temporal tracking could be tested on related tasks such as compressed video denoising.
  • Replacing the contrastive loss with other frequency-sensitive objectives might further reduce artifacts in low-bitrate footage.
  • The motion-guided alignment component could be isolated and applied to non-frequency video enhancement pipelines.

Load-bearing premise

Differentiating frequency subbands spatially and tracking their temporal dynamics will prevent suboptimal results and produce measurable quality gains over prior frequency-based methods.

What would settle it

Running the three public compressed video SR datasets through FCVSR and finding no PSNR improvement or higher complexity than the second-best existing model would falsify the central performance claim.

Figures

Figures reproduced from arXiv: 2502.06431 by Bing Zeng, David Bull, Fan Zhang, Feiyu Chen, Qiang Zhu, Shuyuan Zhu.

Figure 1
Figure 1. Figure 1: Illustration of performance-complexity trade-offs for different com [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: The architecture of the FCVSR model. A compressed LR video is fed into a convolution layer, MGAA, MFFR, and reconstruction (REC) modules [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: The architecture of motion-guided adaptive alignment (MGAA) module. The set of features [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: The architecture of the multi-frequency feature refinement (MFFR) module. [PITH_FULL_IMAGE:figures/full_fig_p006_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Visualization of input feature, output feature, decoupled features, enhanced features in the MFFR module. [PITH_FULL_IMAGE:figures/full_fig_p006_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: The loss functions used for training the FCVSR model. [PITH_FULL_IMAGE:figures/full_fig_p007_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Visual comparison results between FCVSR models and five benchmark methods. [PITH_FULL_IMAGE:figures/full_fig_p010_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: Visualization of subbands of compressed LR frame, Upsampled SR [PITH_FULL_IMAGE:figures/full_fig_p012_8.png] view at source ↗
read the original abstract

Compressed video super-resolution (SR) aims to generate high-resolution (HR) videos from the corresponding low-resolution (LR) compressed videos. Recently, some compressed video SR methods attempt to exploit the spatio-temporal information in the frequency domain, showing great promise in super-resolution performance. However, these methods do not differentiate various frequency subbands spatially or capture the temporal frequency dynamics, potentially leading to suboptimal results. In this paper, we propose a deep frequency-based compressed video SR model (FCVSR) consisting of a motion-guided adaptive alignment (MGAA) network and a multi-frequency feature refinement (MFFR) module. Additionally, a frequency-aware contrastive loss is proposed for training FCVSR, in order to reconstruct finer spatial details. The proposed model has been evaluated on three public compressed video super-resolution datasets, with results demonstrating its effectiveness when compared to existing works in terms of super-resolution performance (up to a 0.14dB gain in PSNR over the second-best model) and complexity.

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 / 2 minor

Summary. The paper proposes FCVSR, a deep frequency-based model for compressed video super-resolution consisting of a motion-guided adaptive alignment (MGAA) network, a multi-frequency feature refinement (MFFR) module, and a frequency-aware contrastive loss. It claims that explicitly differentiating frequency subbands spatially and capturing temporal frequency dynamics yields up to 0.14 dB PSNR gains over prior methods on three public datasets while also improving complexity.

Significance. If the reported gains can be isolated to the frequency-aware components, the work would provide a concrete advance in handling compressed video SR by addressing a stated limitation of prior frequency-domain methods. The evaluation on public datasets and reported complexity improvements are positive features.

major comments (2)
  1. [§4] §4 (Experiments): The paper reports up to 0.14 dB PSNR improvement but provides no component-wise ablations that isolate the contribution of spatial frequency subband differentiation within MFFR or temporal frequency dynamics within MGAA (e.g., MFFR with vs. without per-subband spatial processing). Without these controls, the central attribution of the observed delta to the frequency-aware design choices cannot be verified.
  2. [§3.2] §3.2 (MFFR module): The description of multi-frequency feature refinement does not include quantitative analysis showing that the per-subband spatial differentiation produces measurable refinement gains independent of the overall network capacity or the contrastive loss.
minor comments (2)
  1. [Abstract] Abstract and §1: The claim that prior methods 'do not differentiate various frequency subbands spatially' would benefit from a brief citation or table contrasting the exact architectural choices in the referenced works.
  2. [§4] §4: Table reporting PSNR/SSIM should include standard deviations or multiple runs to support the 0.14 dB claim as statistically meaningful.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive comments highlighting the need for stronger isolation of our frequency-aware components. We address each point below and will revise the manuscript to incorporate additional ablations and analysis.

read point-by-point responses

  1. Referee: [§4] §4 (Experiments): The paper reports up to 0.14 dB PSNR improvement but provides no component-wise ablations that isolate the contribution of spatial frequency subband differentiation within MFFR or temporal frequency dynamics within MGAA (e.g., MFFR with vs. without per-subband spatial processing). Without these controls, the central attribution of the observed delta to the frequency-aware design choices cannot be verified.

    Authors: We agree that the current experiments do not include the requested component-wise controls (e.g., MFFR with vs. without per-subband spatial processing, or MGAA variants isolating temporal frequency dynamics). The reported gains are shown via full-model comparisons against prior methods, but these do not fully isolate the frequency subband contributions from overall capacity or the contrastive loss. We will add the suggested ablation studies in the revised manuscript, including quantitative results for the isolated components, to directly verify their contributions to the 0.14 dB PSNR delta. revision: yes

  2. Referee: [§3.2] §3.2 (MFFR module): The description of multi-frequency feature refinement does not include quantitative analysis showing that the per-subband spatial differentiation produces measurable refinement gains independent of the overall network capacity or the contrastive loss.

    Authors: The MFFR description in §3.2 emphasizes the architectural motivation for per-subband spatial differentiation to address limitations of prior frequency-domain methods. However, we acknowledge the absence of quantitative controls isolating these gains from network capacity or the contrastive loss. In revision, we will supplement the section with controlled experiments (e.g., MFFR variants with/without subband-specific processing) reporting independent refinement metrics to demonstrate the measurable contribution. revision: yes

Circularity Check

0 steps flagged

No circularity: empirical architecture evaluated on public benchmarks

full rationale

The paper proposes a neural architecture (MGAA + MFFR + frequency-aware contrastive loss) for compressed video SR and reports PSNR gains on three public datasets. No derivation chain, first-principles predictions, or fitted quantities are present; performance claims rest on external empirical evaluation rather than any self-definition, self-citation load-bearing step, or renaming of known results. The method is self-contained against standard benchmarks.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

The central claim rests on the empirical effectiveness of the proposed frequency-aware modules and loss function; no explicit free parameters, axioms, or invented entities are detailed in the abstract beyond standard neural network assumptions.

pith-pipeline@v0.9.0 · 5715 in / 1058 out tokens · 35751 ms · 2026-05-23T03:46:52.006491+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

77 extracted references · 77 canonical work pages · 1 internal anchor

  1. [1]

    Sac-rsm: A high-performance uav-side road surveillance model based on super-resolution assisted learning,

    W. Zheng, H. Xu, P. Li, R. Wang, and X. Shao, “Sac-rsm: A high-performance uav-side road surveillance model based on super-resolution assisted learning,”IEEE Internet of Things Journal, 2024

    work page 2024

  2. [2]

    Human face super-resolution on poor quality surveillance video footage,

    M. Farooq, M. N. Dailey, A. Mahmood, J. Moonrinta, and M. Ekpanyapong, “Human face super-resolution on poor quality surveillance video footage,”Neural Computing and Applica- tions, vol. 33, pp. 13 505–13 523, 2021

    work page 2021

  3. [3]

    Cunerf: Cube-based neural radiance field for zero-shot medical image arbitrary-scale super resolution,

    Z. Chen, L. Yang, J.-H. Lai, and X. Xie, “Cunerf: Cube-based neural radiance field for zero-shot medical image arbitrary-scale super resolution,” inProceedings of the IEEE/CVF International Conference on Computer Vision, 2023, pp. 21 185–21 195

    work page 2023

  4. [4]

    Rethink- ing dual-stream super-resolution semantic learning in medical image segmentation,

    Z. Qiu, Y . Hu, X. Chen, D. Zeng, Q. Hu, and J. Liu, “Rethink- ing dual-stream super-resolution semantic learning in medical image segmentation,”IEEE Transactions on Pattern Analysis and Machine Intelligence, 2023

    work page 2023

  5. [5]

    A super-resolution-based feature map compression for machine-oriented video coding,

    J.-H. Kang, M. S. Ali, H.-W. Jeong, C.-K. Choi, Y . Kim, S. Y . Jeong, S.-H. Bae, and H. Y . Kim, “A super-resolution-based feature map compression for machine-oriented video coding,” IEEE Access, vol. 11, pp. 34 198–34 209, 2023

    work page 2023

  6. [6]

    Luma-only resampling-based video coding with cnn-based super resolu- tion,

    C. Lin, Y . Li, J. Li, K. Zhang, and L. Zhang, “Luma-only resampling-based video coding with cnn-based super resolu- tion,” in2023 IEEE International Conference on Visual Com- munications and Image Processing, 2023, pp. 1–5

    work page 2023

  7. [7]

    Optical flow estimation using a spatial pyramid network,

    A. Ranjan and M. J. Black, “Optical flow estimation using a spatial pyramid network,” inProceedings of the IEEE Confer- ence on Computer Vision and Pattern Recognition, 2017, pp. 4161–4170

    work page 2017

  8. [8]

    Raft: Recurrent all-pairs field transforms for optical flow,

    Z. Teed and J. Deng, “Raft: Recurrent all-pairs field transforms for optical flow,” inProceedings of the European Conference on Computer Vision, 2020, pp. 402–419

    work page 2020

  9. [9]

    Deformable convolutional networks,

    J. Dai, H. Qi, Y . Xiong, Y . Li, G. Zhang, H. Hu, and Y . Wei, “Deformable convolutional networks,” inProceedings of the IEEE International Conference on Computer Vision, 2017, pp. 764–773

    work page 2017

  10. [10]

    Deformable convnets v2: More deformable, better results,

    X. Zhu, H. Hu, S. Lin, and J. Dai, “Deformable convnets v2: More deformable, better results,” inProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recog- nition, 2019, pp. 9308–9316

    work page 2019

  11. [11]

    Vivit: A video vision transformer,

    A. Arnab, M. Dehghani, G. Heigold, C. Sun, M. Lu ˇci´c, and C. Schmid, “Vivit: A video vision transformer,” inProceedings of the IEEE/CVF International Conference on Computer Vision, 2021, pp. 6836–6846

    work page 2021

  12. [12]

    Video diffusion models,

    J. Ho, T. Salimans, A. Gritsenko, W. Chan, M. Norouzi, and D. J. Fleet, “Video diffusion models,”Advances in Neural Information Processing Systems, vol. 35, pp. 8633–8646, 2022

    work page 2022

  13. [13]

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

    K. C. Chan, X. Wang, K. Yu, C. Dong, and C. C. Loy, “Basicvsr: The search for essential components in video super-resolution and beyond,” inProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2021, pp. 4947– 4956

    work page 2021

  14. [14]

    Temporal consis- tency learning of inter-frames for video super-resolution,

    M. Liu, S. Jin, C. Yao, C. Lin, and Y . Zhao, “Temporal consis- tency learning of inter-frames for video super-resolution,”IEEE Transactions on Circuits and Systems for Video Technology, vol. 33, no. 4, pp. 1507–1520, 2022

    work page 2022

  15. [15]

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

    Y . Tian, Y . Zhang, Y . Fu, and C. Xu, “Tdan: Temporally- deformable alignment network for video super-resolution,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2020, pp. 3360–3369

    work page 2020

  16. [16]

    Edvr: Video restoration with enhanced deformable convolu- tional networks,

    X. Wang, K. C. Chan, K. Yu, C. Dong, and C. Change Loy, “Edvr: Video restoration with enhanced deformable convolu- tional networks,” inProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops, 2019, pp. 1954–1963

    work page 2019

  17. [17]

    Learning trajectory- aware transformer for video super-resolution,

    C. Liu, H. Yang, J. Fu, and X. Qian, “Learning trajectory- aware transformer for video super-resolution,” inProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2022, pp. 5687–5696

    work page 2022

  18. [18]

    Learning degradation-robust spatiotemporal frequency-transformer for video super-resolution,

    Z. Qiu, H. Yang, J. Fu, D. Liu, C. Xu, and D. Fu, “Learning degradation-robust spatiotemporal frequency-transformer for video super-resolution,”IEEE Transactions on Pattern Analysis and Machine Intelligence, pp. 14 888–14 904, 2023

    work page 2023

  19. [19]

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

    S. Zhou, P. Yang, J. Wang, Y . Luo, and C. C. 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, 2024, pp. 2535– 2545

    work page 2024

  20. [20]

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

    X. Yang, C. He, J. Ma, and L. Zhang, “Motion-guided latent diffusion for temporally consistent real-world video super- resolution,” inEuropean Conference on Computer Vision. Springer, 2025, pp. 224–242

    work page 2025

  21. [21]

    Video compression based on spatio-temporal resolution adaptation,

    M. Afonso, F. Zhang, and D. R. Bull, “Video compression based on spatio-temporal resolution adaptation,”IEEE Transactions on JOURNAL OF LATEX CLASS FILES, VOL. 14, NO. 8, AUGUST 2021 13 Circuits and Systems for Video Technology, vol. 29, no. 1, pp. 275–280, 2018

    work page 2021

  22. [22]

    Down-sampling based video coding using super-resolution technique,

    M. Shen, P. Xue, and C. Wang, “Down-sampling based video coding using super-resolution technique,”IEEE Transactions on Circuits and Systems for Video Technology, vol. 21, no. 6, pp. 755–765, 2011

    work page 2011

  23. [23]

    Efficient video compression via content-adaptive super-resolution,

    M. Khani, V . Sivaraman, and M. Alizadeh, “Efficient video compression via content-adaptive super-resolution,” inProceed- ings of the IEEE/CVF International Conference on Computer Vision, 2021, pp. 4521–4530

    work page 2021

  24. [24]

    Learned low bitrate video compression with space-time super- resolution,

    J. Yang, C. Yang, F. Xiong, F. Wang, and R. Wang, “Learned low bitrate video compression with space-time super- resolution,” inProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2022, pp. 1786– 1790

    work page 2022

  25. [25]

    Bull and F

    D. Bull and F. Zhang,Intelligent image and video compression: communicating pictures. Academic Press, 2021

    work page 2021

  26. [26]

    Blind quality enhancement for compressed video,

    Q. Ding, L. Shen, L. Yu, H. Yang, and M. Xu, “Blind quality enhancement for compressed video,”IEEE Transactions on Multimedia, pp. 5782–5794, 2023

    work page 2023

  27. [27]

    Video compression artifacts removal with spatial-temporal attention- guided enhancement,

    N. Jiang, W. Chen, J. Lin, T. Zhao, and C.-W. Lin, “Video compression artifacts removal with spatial-temporal attention- guided enhancement,”IEEE Transactions on Multimedia, pp. 5657–5669, 2023

    work page 2023

  28. [28]

    Spatio-temporal detail information retrieval for compressed video quality en- hancement,

    D. Luo, M. Ye, S. Li, C. Zhu, and X. Li, “Spatio-temporal detail information retrieval for compressed video quality en- hancement,”IEEE Transactions on Multimedia, vol. 25, pp. 6808–6820, 2022

    work page 2022

  29. [29]

    Comisr: Compression-informed video super-resolution,

    Y . Li, P. Jin, F. Yang, C. Liu, M.-H. Yang, and P. Milan- far, “Comisr: Compression-informed video super-resolution,” inProceedings of the IEEE/CVF International Conference on Computer Vision, 2021, pp. 2543–2552

    work page 2021

  30. [30]

    Compressed domain deep video super-resolution,

    P. Chen, W. Yang, M. Wang, L. Sun, K. Hu, and S. Wang, “Compressed domain deep video super-resolution,”IEEE Trans- actions on Image Processing, vol. 30, pp. 7156–7169, 2021

    work page 2021

  31. [31]

    A codec information assisted framework for efficient compressed video super-resolution,

    H. Zhang, X. Zou, J. Guo, Y . Yan, R. Xie, and L. Song, “A codec information assisted framework for efficient compressed video super-resolution,” inEuropean Conference on Computer Vision, 2022, pp. 220–235

    work page 2022

  32. [32]

    Compression-aware video super-resolution,

    Y . Wang, T. Isobe, X. Jia, X. Tao, H. Lu, and Y .-W. Tai, “Compression-aware video super-resolution,” inProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2023, pp. 2012–2021

    work page 2023

  33. [33]

    Deep compressed video super-resolution with guidance of coding priors,

    Q. Zhu, F. Chen, Y . Liu, S. Zhu, and B. Zeng, “Deep compressed video super-resolution with guidance of coding priors,”IEEE Transactions on Broadcasting, 2024

    work page 2024

  34. [34]

    Fm- vsr: Feature multiplexing video super-resolution for compressed video,

    G. He, S. Wu, S. Pei, L. Xu, C. Wu, K. Xu, and Y . Li, “Fm- vsr: Feature multiplexing video super-resolution for compressed video,”IEEE Access, vol. 9, pp. 88 060–88 068, 2021

    work page 2021

  35. [35]

    Aim 2024 challenge on efficient video super-resolution for av1 compressed content,

    M. V . Conde, Z. Lei, W. Li, C. Bampis, I. Katsavouni- dis, and R. Timofte, “Aim 2024 challenge on efficient video super-resolution for av1 compressed content,”arXiv preprint arXiv:2409.17256, 2024

    work page arXiv 2024

  36. [36]

    Gaussian mask guided attention for compressed video super resolution,

    L. Chen, “Gaussian mask guided attention for compressed video super resolution,” inIEEE 2023 20th International Computer Conference on Wavelet Active Media Technology and Informa- tion Processing, 2023, pp. 1–6

    work page 2023

  37. [37]

    Learning spatiotemporal frequency-transformer for compressed video super-resolution,

    Z. Qiu, H. Yang, J. Fu, and D. Fu, “Learning spatiotemporal frequency-transformer for compressed video super-resolution,” inEuropean Conference on Computer Vision. Springer, 2022, pp. 257–273

    work page 2022

  38. [38]

    Compressed video super- resolution based on hierarchical encoding,

    Y . Jiang, S. Teng, Q. Zhu, C. Feng, C. Zeng, F. Zhang, S. Zhu, B. Zeng, and D. Bull, “Compressed video super- resolution based on hierarchical encoding,”arXiv preprint arXiv:2506.14381, 2025

    work page arXiv 2025

  39. [39]

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

    K. C. Chan, S. Zhou, X. Xu, and C. C. Loy, “Basicvsr++: Improving video super-resolution with enhanced propagation and alignment,” inProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2022, pp. 5972– 5981

    work page 2022

  40. [40]

    Towards progressive multi-frequency representation for image warping,

    J. Xiao, Z. Lyu, C. Zhang, Y . Ju, C. Shui, and K.-M. Lam, “Towards progressive multi-frequency representation for image warping,” inProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2024, pp. 2995– 3004

    work page 2024

  41. [41]

    Multi-frequency representation enhancement with privilege information for video super-resolution,

    F. Li, L. Zhang, Z. Liu, J. Lei, and Z. Li, “Multi-frequency representation enhancement with privilege information for video super-resolution,” inProceedings of the IEEE/CVF Interna- tional Conference on Computer Vision, 2023, pp. 12 814– 12 825

    work page 2023

  42. [42]

    Online video super-resolution with convolu- tional kernel bypass grafts,

    J. Xiao, X. Jiang, N. Zheng, H. Yang, Y . Yang, Y . Yang, D. Li, and K.-M. Lam, “Online video super-resolution with convolu- tional kernel bypass grafts,”IEEE Transactions on Multimedia, vol. 25, pp. 8972–8987, 2023

    work page 2023

  43. [43]

    Fffn: Frame-by-frame feedback fusion network for video super-resolution,

    J. Zhu, Q. Zhang, L. Fei, R. Cai, Y . Xie, B. Sheng, and X. Yang, “Fffn: Frame-by-frame feedback fusion network for video super-resolution,”IEEE Transactions on Multimedia, vol. 25, pp. 6821–6835, 2022

    work page 2022

  44. [44]

    Omni- directional video super-resolution using deep learning,

    A. A. Baniya, T.-K. Lee, P. W. Eklund, and S. Aryal, “Omni- directional video super-resolution using deep learning,”IEEE Transactions on Multimedia, vol. 26, pp. 540–554, 2023

    work page 2023

  45. [45]

    Video super-resolution transformer with masked inter&intra- frame attention,

    X. Zhou, L. Zhang, X. Zhao, K. Wang, L. Li, and S. Gu, “Video super-resolution transformer with masked inter&intra- frame attention,” inProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2024, pp. 25 399– 25 408

    work page 2024

  46. [46]

    En- hancing video super-resolution via implicit resampling-based alignment,

    K. Xu, Z. Yu, X. Wang, M. B. Mi, and A. Yao, “En- hancing video super-resolution via implicit resampling-based alignment,” inProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2024, pp. 2546– 2555

    work page 2024

  47. [47]

    Trajectory-aware shifted state space models for online video super-resolution,

    Q. Zhu, X. Meng, Y . Jiang, F. Zhang, D. Bull, S. Zhu, and B. Zeng, “Trajectory-aware shifted state space models for online video super-resolution,”arXiv preprint arXiv:2508.10453, 2025

    work page arXiv 2025

  48. [48]

    On bayesian adaptive video super reso- lution,

    C. Liu and D. Sun, “On bayesian adaptive video super reso- lution,”IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 36, no. 2, pp. 346–360, 2013

    work page 2013

  49. [49]

    Robust web image/video super- resolution,

    Z. Xiong, X. Sun, and F. Wu, “Robust web image/video super- resolution,”IEEE Transactions on Image Processing, vol. 19, no. 8, pp. 2017–2028, 2010

    work page 2017

  50. [50]

    Dvsrnet: Deep video super-resolution based on pro- gressive deformable alignment and temporal-sparse enhance- ment,

    Q. Zhu, F. Chen, S. Zhu, Y . Liu, X. Zhou, R. Xiong, and B. Zeng, “Dvsrnet: Deep video super-resolution based on pro- gressive deformable alignment and temporal-sparse enhance- ment,”IEEE Transactions on Neural Networks and Learning Systems, 2024

    work page 2024

  51. [51]

    Video super-resolution with pyramid flow-guided deformable alignment network,

    T. Qing, X. Ying, Z. Sha, and J. Wu, “Video super-resolution with pyramid flow-guided deformable alignment network,” in IEEE 2023 3rd International Conference on Electrical Engi- neering and Mechatronics Technology, 2023, pp. 758–764

    work page 2023

  52. [52]

    Ctvsr: Collaborative spatial-temporal transformer for video super- resolution,

    J. Tang, C. Lu, Z. Liu, J. Li, H. Dai, and Y . Ding, “Ctvsr: Collaborative spatial-temporal transformer for video super- resolution,”IEEE Transactions on Circuits and Systems for Video Technology, 2023

    work page 2023

  53. [53]

    Lamd: Latent motion diffusion for video generation,

    Y . Hu, Z. Chen, and C. Luo, “Lamd: Latent motion diffusion for video generation,”arXiv preprint arXiv:2304.11603, 2023

    work page arXiv 2023

  54. [54]

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

    Z. Chen, F. Long, Z. Qiu, T. Yao, W. Zhou, J. Luo, and T. Mei, “Learning spatial adaptation and temporal coherence in diffusion models for video super-resolution,” inProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2024, pp. 9232–9241

    work page 2024

  55. [55]

    Dfvsr: Directional fre- quency video super-resolution via asymmetric and enhancement alignment network

    S. Dong, F. Lu, Z. Wu, and C. Yuan, “Dfvsr: Directional fre- quency video super-resolution via asymmetric and enhancement alignment network.” inProceedings of the International Joint Conferences on Artificial Intelligence, 2023, pp. 681–689

    work page 2023

  56. [56]

    Fast and accurate image super-resolution with deep laplacian pyramid networks,

    W.-S. Lai, J.-B. Huang, N. Ahuja, and M.-H. Yang, “Fast and accurate image super-resolution with deep laplacian pyramid networks,”IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 41, no. 11, pp. 2599–2613, 2018. JOURNAL OF LATEX CLASS FILES, VOL. 14, NO. 8, AUGUST 2021 14

    work page 2018

  57. [57]

    Fourier space losses for efficient perceptual image super-resolution,

    D. Fuoli, L. Van Gool, and R. Timofte, “Fourier space losses for efficient perceptual image super-resolution,” inProceedings of the IEEE/CVF International Conference on Computer Vision, 2021, pp. 2360–2369

    work page 2021

  58. [58]

    Focal frequency loss for image reconstruction and synthesis,

    L. Jiang, B. Dai, W. Wu, and C. C. Loy, “Focal frequency loss for image reconstruction and synthesis,” inProceedings of the IEEE/CVF International Conference on Computer Vision, 2021, pp. 13 919–13 929

    work page 2021

  59. [59]

    Flownet: Learning optical flow with convolutional networks,

    A. Dosovitskiy, P. Fischer, E. Ilg, P. Hausser, C. Hazirbas, V . Golkov, P. Van Der Smagt, D. Cremers, and T. Brox, “Flownet: Learning optical flow with convolutional networks,” inProceedings of the IEEE International Conference on Com- puter Vision, 2015, pp. 2758–2766

    work page 2015

  60. [60]

    Flownet 2.0: Evolution of optical flow estimation with deep networks,

    E. Ilg, N. Mayer, T. Saikia, M. Keuper, A. Dosovitskiy, and T. Brox, “Flownet 2.0: Evolution of optical flow estimation with deep networks,” inProceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2017, pp. 2462– 2470

    work page 2017

  61. [61]

    Video enhancement with task-oriented flow,

    T. Xue, B. Chen, J. Wu, D. Wei, and W. T. Freeman, “Video enhancement with task-oriented flow,”International Journal of Computer Vision, vol. 127, pp. 1106–1125, 2019

    work page 2019

  62. [62]

    Ex- ploring temporal frequency spectrum in deep video deblurring,

    Q. Zhu, M. Zhou, N. Zheng, C. Li, J. Huang, and F. Zhao, “Ex- ploring temporal frequency spectrum in deep video deblurring,” inProceedings of the IEEE/CVF International Conference on Computer Vision, 2023, pp. 12 428–12 437

    work page 2023

  63. [63]

    Image super-resolution using very deep residual channel attention networks,

    Y . Zhang, K. Li, K. Li, L. Wang, B. Zhong, and Y . Fu, “Image super-resolution using very deep residual channel attention networks,” inProceedings of the European Conference on Computer Vision, 2018, pp. 286–301

    work page 2018

  64. [64]

    Motion-adaptive separable collaborative filters for blind motion deblurring,

    C. Liu, X. Wang, X. Xu, R. Tian, S. Li, X. Qian, and M.- H. Yang, “Motion-adaptive separable collaborative filters for blind motion deblurring,” inProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2024, pp. 25 595–25 605

    work page 2024

  65. [65]

    Spatio-temporal filter adaptive network for video deblurring,

    S. Zhou, J. Zhang, J. Pan, H. Xie, W. Zuo, and J. Ren, “Spatio-temporal filter adaptive network for video deblurring,” inProceedings of the IEEE/CVF International Conference on Computer Vision, 2019, pp. 2482–2491

    work page 2019

  66. [66]

    Scale-wise convolution for image restoration,

    Y . Fan, J. Yu, D. Liu, and T. S. Huang, “Scale-wise convolution for image restoration,” inProceedings of the AAAI Conference on Artificial Intelligence, vol. 34, no. 07, 2020, pp. 10 770– 10 777

    work page 2020

  67. [67]

    Efficient multi-scale network with learn- able discrete wavelet transform for blind motion deblurring,

    X. Gao, T. Qiu, X. Zhang, H. Bai, K. Liu, X. Huang, H. Wei, G. Zhang, and H. Liu, “Efficient multi-scale network with learn- able discrete wavelet transform for blind motion deblurring,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2024, pp. 2733–2742

    work page 2024

  68. [68]

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

    S. Nah, S. Baik, S. Hong, G. Moon, S. Son, R. Timofte, and K. Mu Lee, “Ntire 2019 challenge on video deblurring and super-resolution: Dataset and study,” inProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition workshops, 2019, pp. 0–0

    work page 2019

  69. [69]

    Adam: A Method for Stochastic Optimization

    D. P. Kingma, “Adam: A method for stochastic optimization,” arXiv preprint arXiv:1412.6980, 2014

    work page internal anchor Pith review Pith/arXiv arXiv 2014

  70. [70]

    H. 264/avc to hevc video transcoder based on dynamic thresholding and content modeling,

    E. Peixoto, T. Shanableh, and E. Izquierdo, “H. 264/avc to hevc video transcoder based on dynamic thresholding and content modeling,”IEEE Transactions on Circuits and Systems for Video Technology, vol. 24, no. 1, pp. 99–112, 2013

    work page 2013

  71. [71]

    Image quality assessment: from error visibility to structural similarity,

    Z. Wang, A. C. Bovik, H. R. Sheikh, and E. P. Simoncelli, “Image quality assessment: from error visibility to structural similarity,”IEEE Transactions on Image Processing, vol. 13, no. 4, pp. 600–612, 2004

    work page 2004

  72. [72]

    Vmaf: The journey continues,

    Z. Li, C. Bampis, J. Novak, A. Aaron, K. Swanson, A. Moorthy, and J. Cock, “Vmaf: The journey continues,”Netflix Technology Blog, vol. 25, no. 1, 2018

    work page 2018

  73. [73]

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

    Y . Jo, S. W. Oh, J. Kang, and S. J. Kim, “Deep video super- resolution network using dynamic upsampling filters without explicit motion compensation,” inProceedings of the IEEE conference on computer vision and pattern recognition, 2018, pp. 3224–3232

    work page 2018

  74. [74]

    Learning temporal coherence via self-supervision for gan- based video generation,

    M. Chu, Y . Xie, J. Mayer, L. Leal-Taix ´e, and N. Thuerey, “Learning temporal coherence via self-supervision for gan- based video generation,”ACM Transactions on Graphics, vol. 39, no. 4, pp. 75–1, 2020

    work page 2020

  75. [75]

    Detail-revealing deep video super-resolution,

    X. Tao, H. Gao, R. Liao, J. Wang, and J. Jia, “Detail-revealing deep video super-resolution,” inProceedings of the IEEE Inter- national Conference on Computer Vision, 2017, pp. 4472–4480

    work page 2017

  76. [76]

    Video super-resolution with recurrent structure-detail network,

    T. Isobe, X. Jia, S. Gu, S. Li, S. Wang, and Q. Tian, “Video super-resolution with recurrent structure-detail network,” in European Conference on Computer Vision. Springer, 2020, pp. 645–660

    work page 2020

  77. [77]

    Mucan: Multi-correspondence aggregation network for video super- resolution,

    W. Li, X. Tao, T. Guo, L. Qi, J. Lu, and J. Jia, “Mucan: Multi-correspondence aggregation network for video super- resolution,” inEuropean Conference on Computer Vision. Springer, 2020, pp. 335–351

    work page 2020