Pith. sign in

REVIEW 4 major objections 6 minor 5 cited by

Generalized and Efficient 2D Gaussian Splatting for Arbitrary-scale Super-Resolution

T0 review · 4 major / 6 minor · reviewed 2026-08-10 · deepseek-v4-flash

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

Pith's one-line read A feed-forward network that predicts image-conditioned 2D Gaussians and renders them with a scale-aware CUDA rasterizer can outperform implicit neural representations for arbitrary-scale super-resolution in both quality and speed.

desk verdict Solid quality gains and a real high-scale efficiency story, but the paper's speed claim is undercut by a self-contradictory rasterization-ratio setting that must be resolved. read the letter →

arxiv 2501.06838 v5 pith:CDPJUHWP submitted 2025-01-12 eess.IV cs.CV

Du Chen , Liyi Chen , Zhengqiang Zhang , Lei Zhang This is my paper · ORCID
classification eess.IVcs.CV
keywords 2DGaussiansplattingarbitrary-scalesuper-resolutionimplicitneuralrepresentationscale-awarerasterizationcontinuousimagefeed-forwarddecodingwindowattentiondifferentiablerendering
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

The paper tries to establish that 2D Gaussian splatting can be turned into a generalizable, feed-forward model for arbitrary-scale image super-resolution, and that such a model beats implicit neural representation (INR) based methods on both quality and speed. It proposes predicting image-conditioned 2D Gaussians, each with opacity, position, shape, orientation, and color, from the low-resolution image in one forward pass, then rendering the high-resolution output by sampling these continuous Gaussians with a scale-aware GPU rasterizer. If true, this replaces per-pixel MLP queries with region-level primitives that concentrate on textured areas, giving crisper edges and much faster inference, such as 91 milliseconds versus about 540 milliseconds at scaling factor ×12 for a 720×720 output.

What carries the argument

The central object is the 2D Gaussian as an image primitive: each Gaussian contributes α·c·f(x,y) where f is a bivariate normal density with mean μ, standard deviations σ, and correlation ρ, and the image is the sum of all Gaussians. Because this sum is a continuous field, the scaling factor only changes the sampling interval, so arbitrary magnifications are rendered by the same parameter-free rasterizer. The second load-bearing piece is the differentiable CUDA rasterizer that processes Gaussians in parallel and touches only pixels within a local range, controlled by a rasterization ratio r = 0.1, reducing complexity to O(r²s²HWN) while keeping gradients flowing to the decoder.

What would settle it

Render the same natural image after partitioning it into windows of size 12 and as a whole, then compare outputs along the seams: if the tiling assumption fails, visible discontinuities or a measurable PSNR gap should appear. A second check is to run GSASR on LR images whose width and height are not multiples of 12 and compare PSNR against divisible sizes.

Watch

Extended reading notes

Core claim

The paper claims that representing an LR image as a sum of N continuous 2D Gaussians, rather than as a field of individually queried pixels, gives a more powerful and cheaper representation for arbitrary-scale super-resolution. Each Gaussian carries opacity, center position, standard deviation, correlation coefficient, and peak RGB color, so it can locally adapt to the shape and orientation of textures. The decoder conditions learnable Gaussian embeddings on LR image features through windowed cross-attention and shifted-window self-attention, then five heads predict the Gaussian properties. A differentiable, GPU-parallel, scale-aware rasterizer samples the continuous field at a density set by the scaling factor, with a local-range truncation that keeps complexity low. On standard benchmarks with EDSR and RDN backbones, the paper reports that GSASR outperforms listed INR methods and the concurrent GaussianSR in PSNR, SSIM, LPIPS, and DISTS across scaling factors from ×2 to ×30, and that its inference is substantially faster at high scales.

Load-bearing premise

The claim that arbitrary image sizes work rests on the untested assumption that duplicating one shared base Gaussian embedding across all windows produces Gaussians that stay consistent across window boundaries.

Editorial extensions

If this is right

  • A single GSASR model, trained with scaling factors sampled from 1.0 to 4.0, reportedly keeps improving over INR baselines on unseen scales up to ×30, so the Gaussian representation itself carries scale generalization.
  • Because the rasterizer is differentiable, the whole pipeline of encoder, Gaussian decoder, and rendering is trained end-to-end with L1 loss, so the Gaussians are optimized directly for the final super-resolved image.
  • At ×12 for a 720×720 output, GSASR reports 91 ms versus about 540 ms for CiaoSR, indicating that windowed attention plus the GPU rasterizer avoids the per-pixel MLP cost that dominates INR models.
  • The learnable center offsets let Gaussians cluster on textured regions while flat areas stay covered by evenly spaced Gaussians, which is the mechanism behind the reported sharpness on edges and windows.

Reading between the lines

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

  • If the window-tiling generalization is shown to be safe, the same Gaussian decoder could be applied to other continuous-coordinate tasks, such as image compression or video frame interpolation, without changing the rendering core.
  • The speed advantage over INR models at high scales suggests that the bottleneck of arbitrary-scale super-resolution is per-pixel MLP sampling rather than the number of Gaussians, so models with fewer, adaptively placed Gaussians may reach similar quality at even larger scales.
  • A testable extension is to replace the fixed window cross-attention with overlap-aware or global attention for a subset of embeddings, preserving locality where it helps while checking whether boundary consistency improves.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 6 minor

Summary. The paper proposes GSASR, a feed-forward 2D Gaussian splatting model for arbitrary-scale super-resolution. An LR image is encoded by a standard SR backbone, Gaussian embeddings are conditioned on the encoded features through window-based cross-attention and Gaussian interaction blocks, and a custom differentiable CUDA rasterizer renders the SR image by sampling the predicted continuous Gaussians at scale-dependent positions. The model is trained with L1 loss on scales sampled from [1,4] and tested on scales up to ×30. The main claims are that GSASR outperforms INR-based ASR methods in quality and is much faster, especially at high scaling factors.

Significance. If substantiated, the paper's central claim is significant: a Gaussian-based feed-forward representation would outperform INR-based arbitrary-scale SR in both fidelity and speed, while scaling efficiently to large upsampling factors. The manuscript's strengths include the breadth of evaluation (multiple datasets, two encoder backbones, four metrics, scales up to ×30), the ablations reported in the supplementary material, the exploration of a larger HAT-L backbone, and the promise of released code and models. The CUDA rasterization is a useful engineering contribution, and the position-distribution visualizations in Figures 6 and 9 give useful insight into why the representation helps on textured regions.

major comments (4)
  1. [§4.2, Table 1 (LSDIR columns)] The sentence 'GSASR outperforms existing methods in terms of both fidelity metrics (PSNR and SSIM) and perceptual quality metrics (LPIPS and DISTS) under scaling factors from ×2 to ×30' is not supported by the LSDIR columns of Table 1. At ×30, CiaoSR has higher PSNR (19.87 vs 19.82) and lower LPIPS (0.7284 vs 0.7368); at ×24 and ×18, CiaoSR also has higher PSNR (20.40 vs 20.38 and 21.18 vs 21.14). Please either restrict the claim to the datasets and metrics where it holds, or provide a statistical or qualitative justification for the small differences at the largest scales.
  2. [§3.1 vs Supplementary §6.3] The rasterization ratio is stated as r = 0.1 in §3.1 ('we set the ratio as r = 0.1'), but Supplementary §6.3 concludes 'we set r = 1.0 cautiously.' Supplementary Table 13 reports r = 1.0 inference times of 9419 ms (×4) and 3285 ms (×8), while main-text Table 2 reports 543 ms and 195 ms for GSASR, exactly matching the r = 0.1 columns. This is an internal contradiction on the load-bearing efficiency claim. Please state unambiguously which ratio the released code uses and make the text, tables, and code consistent; if r = 1.0 is the deployed setting, the reported speed advantage over CiaoSR no longer holds.
  3. [Table 2 and Supplementary Table 11] The introduction's claim that 'GSASR is much faster than state-of-the-art INR-based models' is contradicted by Table 2 at low scales: at ×2 GSASR takes 1573 ms versus 86 ms for LINF, 107 ms for SRNO, 126 ms for LTE, and 454 ms for LIIF; at ×3 GSASR takes 806 ms versus 89 ms for LINF and 114 ms for SRNO. The same pattern appears in Supplementary Table 11 with the RDN backbone (×2: 1679 ms vs 178 ms for Meta-SR and 202 ms for LINF). The speed advantage is real only against some methods and mostly at larger scales (e.g., versus CiaoSR at ×3/×4 and beyond). Please rephrase the speed claims to state this precisely, or report timings that do not depend on the rasterization ratio.
  4. [§3.2 Gaussian Embedding] The arbitrary-image-size claim relies on duplicating a single learned Ebase across all windows and letting reference positions plus window cross-attention specialize the embeddings. The paper does not analyze whether this tiling produces consistent Gaussians across window boundaries, and training is done on fixed 48×48 patches while testing uses larger images. Please add a quantitative check (e.g., PSNR computed on boundary strips versus image interiors, or a visualization of seams) or at least a discussion of when this tiling assumption could break.
minor comments (6)
  1. [§5] The conclusion contains typos: 'brandly-new' should be 'brand-new' and 'all thoes ablation studies' should be 'all those ablation studies'.
  2. [Supplementary §6.3] The final sentence 'we set r = 1.0 cautiously' directly contradicts the main text and the table in the same section; please reword so the chosen default is clear.
  3. [Supplementary Table 8 caption] The caption lists 'scaling factors × 2, × 3, × 4, × 8, × 8'; the second ×8 appears to be a duplicate, probably intended as ×6 or another scale, and the table rows should be checked against the caption.
  4. [Supplementary Table 11] In the ×4 row for RDN, the LPIPS value '23.88' appears to be a typo for '0.2388' (or a similarly formatted value).
  5. [Figure 2 caption] The caption says 'differential rasterization' where 'differentiable rasterization' is meant.
  6. [Reference [37]] The reference title contains 'manga eetrieval' and should read 'manga retrieval'.

Circularity Check

0 steps flagged · score 0.0 of 10

No circular derivation: GSASR's arbitrary-scale outputs are genuine extrapolations from end-to-end training; the rasterization-ratio inconsistency is a reproducibility flaw, not circularity.

full rationale

The paper's derivation chain is self-contained. The central claim, that a feed-forward Gaussian decoder trained with L1 loss on scales s in [1,4] generalizes to unseen scales up to x30, is checked against external benchmarks (DIV2K, LSDIR, Set5, etc.) with downloaded official models under a unified protocol; no predicted quantity is defined in terms of the fitted parameters. Hyperparameters (m=16, d=180, k=12, L=6) are chosen by ablation and do not encode the test-scale outcomes. The comparison with GaussianSR, from which Lei Zhang is an author, is experimental and not used to justify any premise of the method. The contradictory statements on rasterization ratio r (main text Section 3.1, 'we set the ratio as r = 0.1', vs Supplementary Section 6.3, 'we set r = 1.0 cautiously') are a serious internal inconsistency that affects the speed claim and reproducibility, but it is not circularity: the timings are empirical measurements, not identities derived from the method's own definitions. No uniqueness theorem, ansatz-by-citation, or fitted-input-called-prediction pattern is present in the derivation.

Assumptions & free parameters 6 free parameters · 3 assumptions · 0 invented entities

The method introduces no new physical entities or exotic axioms; it uses standard deep learning components and a custom differentiable rasterizer. The free parameters are hyperparameters tuned by ablation. The main assumptions are benchmark-domain assumptions typical for super-resolution papers.

free parameters (6)
  • Gaussian density m = 16
    Number of Gaussians per LR pixel (N = m × H × W). Chosen by ablation (Table 12); larger m improves PSNR but increases computation.
  • Embedding dimension d = 180
    Dimension of Gaussian embeddings. Chosen by ablation (Table 14); performance saturates around d = 180.
  • Window size k = 12
    Window size for feature partitioning and attention. Chosen by ablation (Table 15); larger k helps but raises self-attention cost.
  • Number of Gaussian interaction blocks L = 6
    Stacked Swin-style blocks in the decoder. Set by hand; no dedicated ablation reported in the paper.
  • Rasterization ratio r = 0.1
    Controls the rendering range of each Gaussian. Main text sets r = 0.1; supplement claims r = 1.0. Ablation (Table 13) shows r = 0.1 gives the same PSNR as r = 1.0 with much lower cost.
  • Training scale range = [1.0, 4.0]
    Scaling factors sampled during training. Tests at scales up to 30 are out-of-distribution extrapolations.
assumptions (3)
  • domain assumption Bicubic downsampling is the correct degradation model for generating LR/HR training pairs.
    The paper follows LIIF and uses bicubic downsampling to create LR patches (Section 4.1). This is standard for benchmark SR but may not hold for real-world degradations.
  • standard math Gradients through the custom CUDA rasterizer are numerically correct.
    The rasterization is described in Algorithm 1 and implemented in CUDA, but no formal verification or numerical gradient check is provided.
  • domain assumption Training on DIV2K with random scales in [1,4] transfers to other benchmarks and scales up to 30.
    The paper evaluates on Set5, Set14, Urban100, etc., after training only on DIV2K. This transfer assumption is empirically validated but not guaranteed.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Generalized and Efficient 2D Gaussian Splatting for Arbitrary-scale Super-Resolution." pith.science (2026). https://pith.science/paper/CDPJUHWP

@misc{pith2026250106838,
  author       = {Pith},
  title        = {Pith review of: Generalized and Efficient 2D Gaussian Splatting for Arbitrary-scale Super-Resolution},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/CDPJUHWP}},
  note         = {Machine review of arXiv:2501.06838}
}
read the original abstract

Implicit Neural Representations (INR) have been successfully employed for Arbitrary-scale Super-Resolution (ASR). However, INR-based models need to query the multi-layer perceptron module numerous times and render a pixel in each query, resulting in insufficient representation capability and low computational efficiency. Recently, Gaussian Splatting (GS) has shown its advantages over INR in both visual quality and rendering speed in 3D tasks, which motivates us to explore whether GS can be employed for the ASR task. However, directly applying GS to ASR is exceptionally challenging because the original GS is an optimization-based method through overfitting each single scene, while in ASR we aim to learn a single model that can generalize to different images and scaling factors. We overcome these challenges by developing two novel techniques. Firstly, to generalize GS for ASR, we elaborately design an architecture to predict the corresponding image-conditioned Gaussians of the input low-resolution image in a feed-forward manner. Each Gaussian can fit the shape and direction of an area of complex textures, showing powerful representation capability. Secondly, we implement an efficient differentiable 2D GPU/CUDA-based scale-aware rasterization to render super-resolved images by sampling discrete RGB values from the predicted continuous Gaussians. Via end-to-end training, our optimized network, namely GSASR, can perform ASR for any image and unseen scaling factors. Extensive experiments validate the effectiveness of our proposed method. The code and models are available at https://github.com/ChrisDud0257/GSASR.

Figures

Figures reproduced from arXiv: 2501.06838 by the authors.

Figure 1
Figure 1. Comparison between GSASR and representative ASR methods on DIV2K. GSASR outperforms previous methods in visual quality with competitive inference time across both in-distribution (×2, ×4) and out￾of-distribution (×6, ×12) scaling factors. veloped in the past decade [5–7, 32, 44, 47, 48, 51, 52, 56– 58, 60]. However, most of the existing SR models [10, 16, 30, 31, 46, 61, 63, 65, 66] are designed to operate only unde… view at source ↗
Figure 2
Figure 2. Overview of GSASR. In the training phase, an LR image is fed into the encoder to extract image features, conditioned on [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Demonstration of the sampling process in 2D rasteriza [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figures from the paper (13 more)
Figure 4
Figure 4. Figure 4: Visualization of GSASR and the competing methods under [PITH_FULL_IMAGE:figures/full_fig_p006_4.png]
Figure 5
Figure 5. Figure 5: Visualization of GSASR and the competing methods under [PITH_FULL_IMAGE:figures/full_fig_p006_5.png]
Figure 6
Figure 6. Figure 6: Demonstration of the rich expressiveness of Gaussians for ASR. The position distribution is obtained by setting [PITH_FULL_IMAGE:figures/full_fig_p008_6.png]
Figure 1
Figure 1. Figure 1: Visualization of GSASR and the competing methods under [PITH_FULL_IMAGE:figures/full_fig_p017_1.png]
Figure 2
Figure 2. Figure 2: Visualization of GSASR and other methods under [PITH_FULL_IMAGE:figures/full_fig_p018_2.png]
Figure 3
Figure 3. Figure 3: Visualization of GSASR and other methods under [PITH_FULL_IMAGE:figures/full_fig_p018_3.png]
Figure 4
Figure 4. Figure 4: Visualization of GSASR and other methods under [PITH_FULL_IMAGE:figures/full_fig_p018_4.png]
Figure 5
Figure 5. Figure 5: Visualization of GSASR and other methods under [PITH_FULL_IMAGE:figures/full_fig_p019_5.png]
Figure 6
Figure 6. Figure 6: Visualization of GSASR and other methods under [PITH_FULL_IMAGE:figures/full_fig_p019_6.png]
Figure 7
Figure 7. Figure 7: Visualization of GSASR and other methods under [PITH_FULL_IMAGE:figures/full_fig_p019_7.png]
Figure 8
Figure 8. Figure 8: Visualization of GSASR and other methods under [PITH_FULL_IMAGE:figures/full_fig_p020_8.png]
Figure 9
Figure 9. Figure 9: Demonstration of the rich expressiveness of Gaussians for ASR. The position distribution is obtained by setting [PITH_FULL_IMAGE:figures/full_fig_p021_9.png]
Figure 10
Figure 10. Figure 10: The training loss curves with and without reference position. [PITH_FULL_IMAGE:figures/full_fig_p023_10.png]

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 5 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. 2D Gaussian Splatting with Semantic Alignment for Image Inpainting

    cs.CV 2025-09 conditional novelty 6.0 of 10

    A 2D Gaussian Splatting encoder-rasterization network with DINO-based semantic alignment achieves competitive image inpainting results.

  2. SAT-SKYLINES: 3D Building Generation from Satellite Imagery and Coarse Geometric Priors

    cs.CV 2025-08 conditional novelty 6.0 of 10

    A 3D building generator that combines satellite imagery with coarse geometric priors via cosine interpolation between noise and prior latents, trained on a new 50K building dataset.

  3. Fine-structure Preserved Real-world Image Super-resolution via Transfer VAE Training

    cs.CV 2025-07 conditional novelty 6.0 of 10

    A transfer training scheme converts Stable Diffusion's 8x VAE into a 4x VAE that stays compatible with the pretrained UNet, improving fine-structure preservation in real-world super-resolution at lower FLOPs.

  4. Texture++: Elevating 3D Asset Texture Resolution with a Region-Aware Diffusion Model

    cs.CV 2026-07 conditional novelty 5.0 of 10

    A view-based, mask-guided diffusion pipeline that super-resolves 3D model textures at 4× without gradient optimization.

  5. Efficient Difficulty-Aware Dynamic Routing for Diffusion-Based Real-World Image Super-Resolution

    cs.CV 2026-07 reject novelty 4.0 of 10

    DDR-SR routes each real-world low-resolution image to one of two diffusion experts based on a high-frequency-loss difficulty score, using a low-compression VAE for hard images and a high-compression VAE for easy image...

Reference graph

Works this paper leans on

28 extracted references · 27 canonical work pages · cited by 5 Pith papers

  1. [1]

    Low-complexity single-image super-resolution based on nonnegative neighbor embedding

    Marco Bevilacqua, Aline Roumy, Christine Guillemot, and Marie Line Alberi-Morel. Low-complexity single-image super-resolution based on nonnegative neighbor embedding. In BMVC, pages 135.1–135.10, 2012. 1, 5

    work page 2012

  2. [2]

    Ciaosr: Continuous implicit attention-in-attention network for arbitrary-scale image super-resolution

    Jiezhang Cao, Qin Wang, Y ongqin Xian, Y awei Li, Bingbing Ni, Zhiming Pi, Kai Zhang, Y ulun Zhang, Radu Timofte, and Luc V an Gool. Ciaosr: Continuous implicit attention-in-attention network for arbitrary-scale image super-resolution. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages 1796–1807, 2023. 1, 2

    work page 2023

  3. [3]

    Activating more pixels in image super-resolution transformer

    Xiangyu Chen, Xintao Wang, Jiantao Zhou, Y u Qiao, and Chao Dong. Activating more pixels in image super-resolution transformer. In IEEE Conference on Computer Vision and Pattern Recognition , pages 22367–22377. IEEE, 2023. 13

    work page 2023

  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 IEEE Conference on Computer Vision and Pattern Recognition, pages 8628–8638, 2021. 1, 2

    work page 2021

  5. [5]

    Flashattention: Fast and memory-efficient exact attention with io-awareness

    Tri Dao, Dan Fu, Stefano Ermon, Atri Rudra, and Christopher R ´e. Flashattention: Fast and memory-efficient exact attention with io-awareness. Advances in Neural Information Processing Systems , 35:16344–16359, 2022. 13

    work page 2022

  6. [6]

    Image quality assessment: Unifying structure and texture similarity

    Keyan Ding, Kede Ma, Shiqi Wang, and Eero P Simoncelli. Image quality assessment: Unifying structure and texture similarity. IEEE Transactions on Pattern Analysis and Machine Intelligence , 44(5):2567–2581, 2020. 1

    work page 2020

  7. [7]

    Accelerating the super-resolution convolutional neural network

    Chao Dong, Chen Change Loy, and Xiaoou Tang. Accelerating the super-resolution convolutional neural network. In European Conference on Computer Vision, pages 391–407. Springer, 2016. 1, 4

    work page 2016

  8. [8]

    Latent modulated function for computational optimal continuous image representation

    Zongyao He and Zhi Jin. Latent modulated function for computational optimal continuous image representation. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages 26026–26035, 2024. 1

    work page 2024

Show all 28 references
  1. [9]

    Rotary position embedding for vision transformer

    Byeongho Heo, Song Park, Dongyoon Han, and Sangdoo Y un. Rotary position embedding for vision transformer. In European Conference on Computer Vision, pages 289–305. Springer, 2024. 13

    2024

  2. [10]

    Gaussiansr: High fidelity 2d gaussian splatting for arbitrary-scale image super-resolution

    Jintong Hu, Bin Xia, Bin Chen, Wenming Y ang, and Lei Zhang. Gaussiansr: High fidelity 2d gaussian splatting for arbitrary-scale image super-resolution. In Proceedings of the Association for the Advancement of Artificial Intelligence , 2025. 1, 2, 9

    2025

  3. [11]

    Meta-sr: A magnification-arbitrary network for super-resolution

    Xuecai Hu, Haoyuan Mu, Xiangyu Zhang, Zilei Wang, Tieniu Tan, and Jian Sun. Meta-sr: A magnification-arbitrary network for super-resolution. In IEEE Conference on Computer Vision and Pattern Recognition , pages 1575–1584, 2019. 1

    2019

  4. [12]

    Single image super-resolution from transformed self-exemplars

    Jia-Bin Huang, Abhishek Singh, and Narendra Ahuja. Single image super-resolution from transformed self-exemplars. In IEEE Conference on Computer Vision and Pattern Recognition, pages 5197–5206, 2015. 1, 3

    2015

  5. [13]

    Adam: A method for stochastic optimization

    Diederik P Kingma and Jimmy Ba. Adam: A method for stochastic optimization. arXiv preprint arXiv:1412.6980, 2014. 13

    2014 arXiv

  6. [14]

    Segment anything

    Alexander Kirillov, Eric Mintun, Nikhila Ravi, Hanzi Mao, Chloe Rolland, Laura Gustafson, Tete Xiao, Spencer Whitehead, Alexan- der C Berg, Wan-Y en Lo, et al. Segment anything. In Proceedings of the IEEE/CVF International Conference on Computer Vision , pages 4015–4026, 2023. 13

    2023

  7. [15]

    Local texture estimator for implicit representation function

    Jaewon Lee and Kyong Hwan Jin. Local texture estimator for implicit representation function. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 1929–1938, 2022. 1

    1929

  8. [16]

    Lsdir: A large scale dataset for image restoration

    Y awei Li, Kai Zhang, Jingyun Liang, Jiezhang Cao, Ce Liu, Rui Gong, Y ulun Zhang, Hao Tang, Y un Liu, Denis Demandolx, et al. Lsdir: A large scale dataset for image restoration. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 1775–1...

    2023

  9. [17]

    Enhanced deep residual networks for single image super-resolution

    Bee Lim, Sanghyun Son, Heewon Kim, Seungjun Nah, and Kyoung Mu Lee. Enhanced deep residual networks for single image super-resolution. In IEEE Conference on Computer Vision and Pattern Recognition Workshop , pages 136–144, 2017. 1, 3, 4, 5, 6, 7, 8, 12

    2017

  10. [18]

    Swin transformer: Hierarchical vision transformer using shifted windows

    Ze Liu, Y utong Lin, Y ue Cao, Han Hu, Yixuan Wei, Zheng Zhang, Stephen Lin, and Baining Guo. Swin transformer: Hierarchical vision transformer using shifted windows. In Proceedings of the IEEE/CVF international conference on computer vision , pages 10012–10022, 2021. 13

    2021

  11. [19]

    A database of human segmented natural images and its application to evaluating segmentation algorithms and measuring ecological statistics

    David Martin, Charless Fowlkes, Doron Tal, and Jitendra Malik. A database of human segmented natural images and its application to evaluating segmentation algorithms and measuring ecological statistics. In IEEE International Conference on Computer Vision , pages 416–423. IEEE,...

    2001

  12. [20]

    Sketch-based manga eetrieval using manga109 dataset

    Y usuke Matsui, Kota Ito, Y uji Aramaki, Azuma Fujimoto, Toru Ogawa, Toshihiko Y amasaki, and Kiyoharu Aizawa. Sketch-based manga eetrieval using manga109 dataset. Multimedia Tools and Applications, 76(20):21811–21838, 2017. 1, 3

    2017

  13. [21]

    Ntire 2017 challenge on single image super- resolution: Methods and results

    Radu Timofte, Eirikur Agustsson, Luc V an Gool, Ming-Hsuan Y ang, and Lei Zhang. Ntire 2017 challenge on single image super- resolution: Methods and results. In IEEE Conference on Computer Vision and Pattern Recognition Workshop, pages 114–125, 2017. 1, 2, 7, 12

    2017

  14. [22]

    Image quality assessment: From error visibility to structural similarity

    Zhou Wang, Alan C Bovik, Hamid R Sheikh, and Eero P Simoncelli. Image quality assessment: From error visibility to structural similarity. IEEE Transactions on Image Processing, 13(4):600–612, 2004. 1, 3, 4, 5

    2004

  15. [23]

    Super-resolution neural operator

    Min Wei and Xuesong Zhang. Super-resolution neural operator. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 18247–18256, 2023. 1

    2023

  16. [24]

    Local implicit normalizing flow for arbitrary- scale image super-resolution

    Jie-En Y ao, Li-Y uan Tsao, Yi-Chen Lo, Roy Tseng, Chia-Che Chang, and Chun-Yi Lee. Local implicit normalizing flow for arbitrary- scale image super-resolution. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages 1776–1785, 2023. 1 14

    2023

  17. [25]

    On single image scale-up using sparse-representations

    Roman Zeyde, Michael Elad, and Matan Protter. On single image scale-up using sparse-representations. In International Conference on Curves and Surfaces , pages 711–730. Springer, 2010. 1, 5

    2010

  18. [26]

    The unreasonable effectiveness of deep features as a perceptual metric

    Richard Zhang, Phillip Isola, Alexei A Efros, Eli Shechtman, and Oliver Wang. The unreasonable effectiveness of deep features as a perceptual metric. In IEEE Conference on Computer Vision and Pattern Recognition , pages 586–595, 2018. 1

    2018

  19. [27]

    Image super-resolution using very deep residual channel attention networks

    Y ulun Zhang, Kunpeng Li, Kai Li, Lichen Wang, Bineng Zhong, and Y un Fu. Image super-resolution using very deep residual channel attention networks. In European Conference on Computer Vision, pages 286–301, 2018. 3, 11

    2018

  20. [28]

    Residual dense network for image super-resolution

    Y ulun Zhang, Y apeng Tian, Y u Kong, Bineng Zhong, and Y un Fu. Residual dense network for image super-resolution. In IEEE Conference on Computer Vision and Pattern Recognition, pages 2472–2481, 2018. 1, 2, 4, 5, 7, 8, 9 15

    2018

Pith tools

Reviewed August 10, 2026 · model on record in the stance chip above.