Pith. sign in

REVIEW 4 major objections 5 minor 2 cited by

Semantic Segmentation Prior for Diffusion-Based Real-World Super-Resolution

T0 review · 4 major / 5 minor · reviewed 2026-08-11 · deepseek-v4-flash

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

Pith's one-line read This paper claims that conditioning a diffusion super-resolution model on a pixel-level segmentation mask that is jointly refined by a separate segmentation diffusion model produces more semantically faithful restorations than text-prompt…

desk verdict Genuine dual-diffusion mutual-refinement design, but the semantic-fidelity evaluation is circular and the no-reference gains are mixed, so treat as promising work in progress. read the letter →

arxiv 2412.02960 v1 pith:7IOXLHXN submitted 2024-12-04 cs.CV

Jiahua Xiao , Jiawei Zhang , Dongqing Zou , Xiaodan Zhang , Jimmy Ren , Xing Wei This is my paper · ORCID
classification cs.CV
keywords semanticsegmentationreal-worldsuper-resolutiondiffusionmodelsdual-diffusionframeworkmutualrefinementfidelityStableopen-vocabulary
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

This paper claims that real-world image super-resolution fails to preserve what objects actually are when guided only by text prompts, which can mislabel or miss salient objects and create ambiguous regions where multiple prompts compete. The authors propose adding a semantic segmentation map as a pixel-level control condition, so that every pixel is explicitly assigned a class label. Because segmentation maps on severely degraded images are themselves unreliable, the framework runs two diffusion models side by side—one for super-resolution, one for segmentation—and lets them exchange updated information at each denoising step through a Dual-Modality Bridge. The central claim is that this mutual refinement yields restored images that are both realistic and semantically faithful, outperforming text-prompt-guided methods such as SeeSR and PASD on perceptual quality and semantic-accuracy metrics.

What carries the argument

The central object is the Dual-Modality Bridge (DMB), a module that sits between two diffusion models operating in different modalities—image latents and discrete segmentation masks. At each time step $t$, DMB's ImgAided component extracts updated image features from the super-resolution latent $Z_t$ and feeds them into the segmentation denoiser, while its Seg Controller extracts refined segmentation features from the mask state $S_t$ and feeds them into the image denoiser. This bidirectional information flow is what lets the two tasks mutually refine each other during reverse diffusion; without it, a static segmentation prediction (e.g., from SegFormer) is too inaccurate on heavily degraded inputs to guide restoration.

What would settle it

Run a human study where annotators label semantic errors in the restored images (e.g., hydrant-like textures where a person should be) for SegSR and SeeSR outputs. If SegSR does not show lower human-judged semantic error rates than SeeSR on the same inputs, the paper's central claim that pixel-level segmentation control improves semantic fidelity over text prompts is not supported.

Watch

Extended reading notes

Core claim

SegSR's central discovery is that a semantic segmentation prior can be made useful for diffusion-based super-resolution if it is jointly refined rather than computed once from the degraded input. The super-resolution branch (SRDM), built on StableDiffusion, receives an updated segmentation condition at every reverse step; the segmentation branch (SegDM), a discrete diffusion model, receives updated image information from the super-resolution branch. The Dual-Modality Bridge (DMB) carries this two-way information flow. The paper shows that this joint refinement progressively corrects segmentation errors (e.g., a rock region mislabeled as 'animal') and simultaneously prevents semantic mistakes in the restored image (e.g., a building rendered with mountain texture). As a result, the restored images preserve semantic structures more faithfully than methods that rely on implicit localization through text-prompt cross-attention.

Load-bearing premise

The whole method assumes the pseudo-labels produced by X-Decoder on high-quality images are a correct ground truth for what objects actually are.

Editorial extensions

If this is right

  • Replacing text-prompt conditioning with pixel-level segmentation control removes the two failure modes the paper identifies: missed salient objects and spatial ambiguity from competing prompts.
  • Jointly refining the segmentation mask during diffusion makes high-quality image information available to the segmentation model, so the mask improves as restoration progresses rather than being frozen at the initial degraded estimate.
  • Because the segmentation branch uses a discrete diffusion model, SegSR inherits distribution-level modeling of masks, allowing it to correct even large mislabeled regions rather than merely smoothing an initial prediction.
  • On no-reference quality metrics (MANIQA, CLIPIQA, MUSIQ) and on the semantic-accuracy metric ACC, SegSR reports consistent gains over SeeSR and other diffusion baselines on both synthetic and real-world benchmarks.

Reading between the lines

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

  • A testable extension is to let a user edit the initial segmentation mask before or during denoising; since the mask is an explicit control condition, SegSR would give interactive control over which semantic content to restore, unlike prompt-based methods.
  • Because the mutual-refinement idea is not specific to super-resolution, the same dual-diffusion design could be applied to other inverse problems where a semantic map and a high-quality image are jointly inferable, such as deblurring, dehazing, or inpainting.
  • The paper's semantic-accuracy metric (ACC) is measured with X-Decoder on the output against X-Decoder on the high-quality reference; since X-Decoder also generates the training ground truth, the reported semantic gains may partly reflect agreement with X-Decoder's label space. A human-annotated semantic-error study would tell whether the improvement is perceptual as well as machine-measured.
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 / 5 minor

Summary. The paper proposes SegSR, a real-world super-resolution framework that conditions a StableDiffusion-based super-resolution branch on semantic segmentation masks produced by a discrete diffusion segmentation branch, with a Dual-Modality Bridge (DMB) exchanging information between the two branches at each reverse-diffusion step. The central claim is that adding pixel-level semantic control and mutually refining the segmentation and image restoration avoids the recognition failures and semantic ambiguities of prompt-based methods such as SeeSR and PASD, thereby preserving semantic structures more effectively. Experiments are reported on DIV2K-Val, OST-Val, RealSR, and RealLQ250, with comparisons against eight baselines and an ablation table, plus qualitative figures.

Significance. The idea of jointly refining segmentation and super-resolution within a dual-diffusion framework is timely and potentially useful for the Real-ISR community. The paper provides extensive comparisons and a clear conceptual motivation, and the ablation table shows monotonic ACC improvements when SegFormer, SegDM, and DMB are added. However, the central semantic-preservation claim is currently supported only by an evaluation that is circular with respect to the training labels, and the quantitative comparisons are not consistent across datasets. If validated with an independent semantic-fidelity evaluation, the framework would be a meaningful step beyond text-prompt-only conditioning, but that validation is not yet present.

major comments (4)
  1. [§4.1, §4.3, Table 2] The semantic-preservation claim is evaluated entirely within X-Decoder's pseudo-label space. Section 4.1 states that ground-truth segmentation masks for training are obtained from X-Decoder, and Table 2 defines ACC as the accuracy of X-Decoder predictions on restoration results against X-Decoder predictions on HQ images. SegSR is therefore trained to reproduce X-Decoder's label space, and ACC rewards any method that shares X-Decoder's biases. Exp. (5), the 'upper bound,' also uses X-Decoder on HQ, so the entire semantic argument is contained within one model's predictions. I request an independent semantic evaluation: human-rated semantic faithfulness, human-annotated masks on a sample, or a different open-vocabulary segmentation model that was not used in training.
  2. [Table 1] Table 1 does not support the claim of strong no-reference performance on real-world data. On RealSR, SegSR scores MUSIQ 67.80, MANIQA 0.5233, and CLIPIQA 0.6906, while SeeSR scores 69.82, 0.5407, and 0.6701, and PASD scores 69.09, 0.5088, and 0.5892; SegSR is not consistently better than several baselines on two of the three metrics. The text in Section 4.2 should be tempered or supported by statistical tests across datasets.
  3. [Table 2, §4.3] No error bars or significance tests are reported for the ablations, and the differences supporting the contribution of DMB are small and mixed on several metrics: on DIV2K-Val, MUSIQ decreases from 72.40 in Exp. (3) to 72.29 in Exp. (4), while MANIQA increases from 0.5840 to 0.6006 and ACC from 0.7009 to 0.7116. Diffusion sampling is stochastic, so bootstrapped confidence intervals or multiple-seed results are necessary to establish that the SegDM and DMB improvements are not noise.
  4. [§3.3] The Dual-Modality Bridge is central to the method but is not specified in the main text: Section 3.3 only states that DMB consists of a Seg Controller and an ImgAided component, and defers details to the supplementary material. Without architecture details, input/output dimensions, and the conditioning mechanism, the method cannot be reproduced from the paper. Please move the DMB specification into the main text or provide an appendix within the manuscript.
minor comments (5)
  1. [§1] In the Introduction, 'balance perceptual quality with fidelity perversion' should likely read 'fidelity preservation' or 'fidelity'; as written it is contradictory.
  2. [Figure 5] The caption contains the typo 'lease zoom in for details' (should be 'Please').
  3. [References] Reference [33] is listed as appearing in 'MACCAI'; the correct venue is MICCAI.
  4. [§3.3] The notation for the DMB conditions (I_t, C_t) and the LQ embeddings (Z_lq, F_lq) is introduced only in Figure 3; the text should define these symbols when they first appear in Section 3.3.
  5. [§2] In Related Work, the stray text 'diffusion-2-1' appears after the StableDiffusion footnote; this looks like a leftover fragment and should be removed.

Circularity Check

2 steps flagged · score 5.0 of 10

Semantic-fidelity claim rests on a circular evaluation: the ACC metric uses X-Decoder on HQ images as ground truth, and the same X-Decoder model produced the semantic pseudo-labels used to train SegDM.

  1. self definitional [Section 4.1 (Training Datasets) and Section 4.3 (Ablation Studies, Table 2)]
    "For semantic segmentation training, the ground truth segmentation masks are obtained from the pretrained open vocabulary segmentation model, X-Decoder [55]. ... The ACC metric represents the accuracy of semantic segmentation predictions by X-Decoder on the restoration results from SRDM, using the predictions on HQ images as the ground truth mask."

    The segmentation branch is trained to reproduce X-Decoder pseudo-labels on HQ images, and the only quantitative semantic-fidelity metric, ACC, measures agreement between X-Decoder on the restored image and X-Decoder on the HQ image. Semantic correctness is therefore operationalized as consistency with the same model that generated the training target. If X-Decoder has systematic biases or misses objects, SegDM is optimized to reproduce those errors, and ACC rewards any method that learns X-Decoder's label space rather than true semantic content. No independent segmentation ground truth or human semantic evaluation is reported.

  2. self definitional [Section 4.3, Exp. (5)]
    "In Exp. (5), we use X-decoder [55] to perform segmentation prediction on the HQ images. ... Notably, in both quantitative and qualitative results for semantic segmentation prediction and image restoration, our method closely approximates the performance of Exp. (5)."

    The claimed upper bound is defined by feeding SRDM the X-Decoder mask computed from the HQ image, which is exactly the source of the training pseudo-labels. Approaching Exp. (5) therefore demonstrates that SegSR reproduces X-Decoder's label assignments, not that the restored image is semantically correct in an independent sense. The 'upper bound' is internal to the X-Decoder label space, so it cannot validate the claim that semantic structures are preserved against true semantics.

full rationale

The dual-diffusion architecture and the Dual-Modality Bridge are genuine methodological contributions, and the photorealism metrics (MANIQA, MUSIQ, CLIPIQA) are independent of the circularity. However, the paper's central claim of 'preserving semantic structures more effectively' is quantitatively supported only by the ACC metric in Table 2, and that metric's ground truth is X-Decoder on HQ images while the training pseudo-labels also come from X-Decoder on HQ images. This makes the semantic-fidelity evaluation self-referential: the model is trained to imitate X-Decoder, and then scored on how well it imitates X-Decoder. The Exp. (5) upper bound is similarly contained within the same pseudo-label space. No human study or external segmentation benchmark is used, and the comparison against SOTA methods in Table 1 does not include ACC, leaving the headline semantic advantage supported mainly by selected qualitative examples. This is a real but partial circularity, so the score is 5 rather than higher.

Assumptions & free parameters 1 free parameters · 5 assumptions · 1 invented entities

The central claim rests on the reliability of pseudo-labels from X-Decoder, the validity of the synthetic degradation pipeline, and the sufficiency of the 34-class semantic space. The DMB is the only genuinely new component, and it has direct ablation support. No parameters are fitted to data in a way that undermines the claim, though the class count is a hand-chosen design decision.

free parameters (1)
  • Semantic class set size = 34
    The paper assumes 34 classes for indoor and outdoor scenes, generated by X-Decoder prompts (Section 4.1). The central claim depends on this label space covering all salient objects; if classes are missing, segmentation cannot provide the claimed comprehensive perception.
assumptions (5)
  • domain assumption Pretrained StableDiffusion (v2.1) provides a strong natural image prior suitable for SR refinement.
    SRDM is built on StableSR/StableDiffusion and the method assumes the frozen priors help generate realistic textures. This is common in the field, not proven here.
  • domain assumption The Real-ESRGAN/SeeSR degradation pipeline approximates real-world degradation distributions.
    All training pairs are synthesized with this pipeline (Section 4.1); if real degradations differ, the method may not transfer.
  • ad hoc to paper X-Decoder's semantic pseudo-labels on HQ images are reliable ground truth for segmentation.
    Training masks and the ACC evaluation both come from X-Decoder (Sections 4.1 and 4.3). This is a core, unverified assumption.
  • domain assumption The 34-class label set suffices for the salient semantics in the test images.
    The paper selects 34 classes for indoor and outdoor scenes; any object outside this set cannot be represented in the segmentation prior.
  • standard math Standard diffusion forward and reverse processes as defined in DDPM and multinomial diffusion hold.
    SRDM and SegDM rely on the noise schedules and denoising equations from the cited diffusion works.
invented entities (1)
  • Dual-Modality Bridge (DMB) independent evidence
    purpose: Transfers intermediate image features from SRDM to SegDM and segmentation features from SegDM to SRDM at each reverse step, enabling mutual refinement.
    The DMB is a new module; its effect is isolated in the ablation (Exp. 3 vs Exp. 4 in Table 2), where adding it improves ACC and no-reference metrics. It is not a physics-style entity but an architectural component.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Semantic Segmentation Prior for Diffusion-Based Real-World Super-Resolution." pith.science (2026). https://pith.science/paper/7IOXLHXN

@misc{pith2026241202960,
  author       = {Pith},
  title        = {Pith review of: Semantic Segmentation Prior for Diffusion-Based Real-World Super-Resolution},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/7IOXLHXN}},
  note         = {Machine review of arXiv:2412.02960}
}
read the original abstract

Real-world image super-resolution (Real-ISR) has achieved a remarkable leap by leveraging large-scale text-to-image models, enabling realistic image restoration from given recognition textual prompts. However, these methods sometimes fail to recognize some salient objects, resulting in inaccurate semantic restoration in these regions. Additionally, the same region may have a strong response to more than one prompt and it will lead to semantic ambiguity for image super-resolution. To alleviate the above two issues, in this paper, we propose to consider semantic segmentation as an additional control condition into diffusion-based image super-resolution. Compared to textual prompt conditions, semantic segmentation enables a more comprehensive perception of salient objects within an image by assigning class labels to each pixel. It also mitigates the risks of semantic ambiguities by explicitly allocating objects to their respective spatial regions. In practice, inspired by the fact that image super-resolution and segmentation can benefit each other, we propose SegSR which introduces a dual-diffusion framework to facilitate interaction between the image super-resolution and segmentation diffusion models. Specifically, we develop a Dual-Modality Bridge module to enable updated information flow between these two diffusion models, achieving mutual benefit during the reverse diffusion process. Extensive experiments show that SegSR can generate realistic images while preserving semantic structures more effectively.

Figures

Figures reproduced from arXiv: 2412.02960 by the authors.

Figure 1
Figure 1. Comparison of Real-ISR results between SegSR conditioned on segmentation masks and prompt-guided methods (examplifed [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. Mutual Refinement within SegSR. We present the final result predictions at different steps t of the inverse diffusion process for both SRDM and SegDM. (a) To provide semantic segmentation priors for Real-ISR, the pretrained Segformer [45] predicts segmentation masks from degraded images, but these predictions become inaccurate when the degradation is severe. As a result, the Segformer-guided SRDM struggles to restor… view at source ↗
Figure 3
Figure 3. Overview of SegSR. Framework comprises three key parts: i) SRDM performs super-resolution diffusion process, conditioned on LQ image embeding Zlq and gradually updated segmentation prior St from SegDM to generate high-realness image; ii) SegDM conducts semantic segmentation diffusion process, conditioned on LQ image features Flq and iteratively restored image information Zt from SRDM to improve the accuracy of segme… view at source ↗
Figures from the paper (3 more)
Figure 4
Figure 4. Figure 4: Qualitative comparisons on synthetic benchmarks: DIV2K-Val [ [PITH_FULL_IMAGE:figures/full_fig_p006_4.png]
Figure 5
Figure 5. Figure 5: Qualitative comparisons on real-world benchmarks: RealSR [ [PITH_FULL_IMAGE:figures/full_fig_p007_5.png]
Figure 6
Figure 6. Figure 6: Ablation study to validate the effectiveness of initial Segformer, SegDM, and DMB. The segmentation masks shown below each [PITH_FULL_IMAGE:figures/full_fig_p008_6.png]

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 2 Pith papers

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

  1. Semantic-Guided Cross-Sensor Super Resolution of Remote Sensing Images: A Gated Dual Conditioning Flow Matching Model

    cs.CV 2025-10 conditional novelty 6.0 of 10

    A gated dual-conditioning flow-matching model achieves 10 m→2 m cross-sensor super-resolution with a 38% FID reduction over the best baseline on a rare-landform (retrogressive thaw slump) benchmark.

  2. MicroZoom: Structure-Preserving Detail Synthesis at Extreme Scale

    cs.CV 2026-07 conditional novelty 5.0 of 10

    A cascaded, segmentation-conditioned, per-instance diffusion method synthesizes globally coherent gigapixel microscopic detail from a phone photo and sparse microscope references at up to 350×.

Reference graph

Works this paper leans on

56 extracted references · 43 canonical work pages · cited by 2 Pith papers

  1. [1]

    Semantic segmentation guided real-world super-resolution

    Andreas Aakerberg, Anders S Johansen, Kamal Nasrollahi, and Thomas B Moeslund. Semantic segmentation guided real-world super-resolution. In CVPR, 2022. 3

    work page 2022

  2. [2]

    Ntire 2017 challenge on single image super-resolution: Dataset and study

    Eirikur Agustsson and Radu Timofte. Ntire 2017 challenge on single image super-resolution: Dataset and study. In CVPRW, 2017. 6

    work page 2017

  3. [3]

    Dream- clear: High-capacity real-world image restoration with privacy-safe dataset curation

    Yuang Ai, Xiaoqiang Zhou, Huaibo Huang, Xiaotian Han, Zhengyu Chen, Quanzeng You, and Hongxia Yang. Dream- clear: High-capacity real-world image restoration with privacy-safe dataset curation. In NIPS, 2024. 6, 7

    work page 2024

  4. [4]

    Toward real-world single image super-resolution: A new benchmark and a new model

    Jianrui Cai, Hui Zeng, Hongwei Yong, Zisheng Cao, and Lei Zhang. Toward real-world single image super-resolution: A new benchmark and a new model. In ICCV, 2019. 6, 7, 8

    work page 2019

  5. [5]

    Activating more pixels in image super- resolution transformer

    Xiangyu Chen, Xintao Wang, Jiantao Zhou, Yu Qiao, and Chao Dong. Activating more pixels in image super- resolution transformer. In CVPR, 2023. 1

    work page 2023

  6. [6]

    Diffusion models beat gans on image synthesis

    Prafulla Dhariwal and Alexander Nichol. Diffusion models beat gans on image synthesis. NIPS, 2021. 1, 2

    work page 2021

  7. [7]

    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. TPAMI, 2020. 6

    work page 2020

  8. [8]

    Image super-resolution using deep convolutional net- works

    Chao Dong, Chen Change Loy, Kaiming He, and Xiaoou Tang. Image super-resolution using deep convolutional net- works. TPAMI, 2015. 1

    work page 2015

Show all 56 references
  1. [9]

    Adadiffsr: Adaptive region-aware dynamic acceleration diffusion model for real-world image super-resolution

    Yuanting Fan, Chengxu Liu, Nengzhong Yin, Changlong Gao, and Xueming Qian. Adadiffsr: Adaptive region-aware dynamic acceleration diffusion model for real-world image super-resolution. ECCV, 2024. 1

    2024

  2. [10]

    Generative adversarial networks

    Ian Goodfellow, Jean Pouget-Abadie, Mehdi Mirza, Bing Xu, David Warde-Farley, Sherjil Ozair, Aaron Courville, and Yoshua Bengio. Generative adversarial networks. CACM,

  3. [11]

    Div8k: Diverse 8k resolution image dataset

    Shuhang Gu, Andreas Lugmayr, Martin Danelljan, Manuel Fritsche, Julien Lamour, and Radu Timofte. Div8k: Diverse 8k resolution image dataset. In ICCVW, 2019. 6

    2019

  4. [12]

    Segnext: Rethinking convolutional attention design for semantic segmentation

    Meng-Hao Guo, Cheng-Ze Lu, Qibin Hou, Zhengning Liu, Ming-Ming Cheng, and Shi-Min Hu. Segnext: Rethinking convolutional attention design for semantic segmentation. NIPS, 2022. 2

    2022

  5. [13]

    Denoising diffu- sion probabilistic models

    Jonathan Ho, Ajay Jain, and Pieter Abbeel. Denoising diffu- sion probabilistic models. NIPS, 2020. 1, 2

    2020

  6. [14]

    Argmax flows and multinomial dif- fusion: Learning categorical distributions

    Emiel Hoogeboom, Didrik Nielsen, Priyank Jaini, Patrick Forr´e, and Max Welling. Argmax flows and multinomial dif- fusion: Learning categorical distributions. NIPS, 2021. 2

    2021

  7. [15]

    A style-based generator architecture for generative adversarial networks

    Tero Karras, Samuli Laine, and Timo Aila. A style-based generator architecture for generative adversarial networks. In CVPR, 2019. 6

    2019

  8. [16]

    Imagic: Text-based real image editing with diffusion models

    Bahjat Kawar, Shiran Zada, Oran Lang, Omer Tov, Huiwen Chang, Tali Dekel, Inbar Mosseri, and Michal Irani. Imagic: Text-based real image editing with diffusion models. In CVPR, 2023. 2

    2023

  9. [17]

    Musiq: Multi-scale image quality transformer

    Junjie Ke, Qifei Wang, Yilin Wang, Peyman Milanfar, and Feng Yang. Musiq: Multi-scale image quality transformer. In ICCV, 2021. 6

    2021

  10. [18]

    Accurate image super-resolution using very deep convolutional net- works

    Jiwon Kim, Jung Kwon Lee, and Kyoung Mu Lee. Accurate image super-resolution using very deep convolutional net- works. In CVPR, 2016. 1

    2016

  11. [19]

    Segment any- thing

    Alexander Kirillov, Eric Mintun, Nikhila Ravi, Hanzi Mao, Chloe Rolland, Laura Gustafson, Tete Xiao, Spencer White- head, Alexander C Berg, Wan-Yen Lo, et al. Segment any- thing. In ICCV, 2023. 3

    2023

  12. [20]

    Denoising diffu- sion semantic segmentation with mask prior modeling.arXiv preprint arXiv:2306.01721, 2023

    Zeqiang Lai, Yuchen Duan, Jifeng Dai, Ziheng Li, Ying Fu, Hongsheng Li, Yu Qiao, and Wenhai Wang. Denoising diffu- sion semantic segmentation with mask prior modeling.arXiv preprint arXiv:2306.01721, 2023. 2, 5

    2023 arXiv

  13. [21]

    Srdiff: Single image super-resolution with diffusion probabilistic models

    Haoying Li, Yifan Yang, Meng Chang, Shiqi Chen, Huajun Feng, Zhihai Xu, Qi Li, and Yueting Chen. Srdiff: Single image super-resolution with diffusion probabilistic models. Neurocomputing, 2022. 2

    2022

  14. [22]

    Blip: Bootstrapping language-image pre-training for unified vision-language understanding and generation

    Junnan Li, Dongxu Li, Caiming Xiong, and Steven Hoi. Blip: Bootstrapping language-image pre-training for unified vision-language understanding and generation. In ICCV,

  15. [23]

    Blip-2: Bootstrapping language-image pre-training with frozen image encoders and large language models

    Junnan Li, Dongxu Li, Silvio Savarese, and Steven Hoi. Blip-2: Bootstrapping language-image pre-training with frozen image encoders and large language models. In ICML,

  16. [24]

    Efficient and degradation-adaptive network for real-world image super- resolution

    Jie Liang, Hui Zeng, and Lei Zhang. Efficient and degradation-adaptive network for real-world image super- resolution. In ECCV, 2022. 6, 7

    2022

  17. [25]

    Efficient and degradation-adaptive network for real-world image super- resolution

    Jie Liang, Hui Zeng, and Lei Zhang. Efficient and degradation-adaptive network for real-world image super- resolution. In ECCV, 2022. 1

    2022

  18. [26]

    Diff- bir: Towards blind image restoration with generative diffu- sion prior

    Xinqi Lin, Jingwen He, Ziyan Chen, Zhaoyang Lyu, Bo Dai, Fanghua Yu, Wanli Ouyang, Yu Qiao, and Chao Dong. Diff- bir: Towards blind image restoration with generative diffu- sion prior. arXiv preprint arXiv:2308.15070, 2023. 1, 2, 6, 7

    2023 arXiv

  19. [27]

    Visual instruction tuning

    Haotian Liu, Chunyuan Li, Qingyang Wu, and Yong Jae Lee. Visual instruction tuning. NIPS, 36, 2024. 2

    2024

  20. [28]

    Transformer for single image super-resolution

    Zhisheng Lu, Juncheng Li, Hong Liu, Chaoyan Huang, Lin- lin Zhang, and Tieyong Zeng. Transformer for single image super-resolution. In CVPR, 2022. 1

    2022

  21. [29]

    Content-aware local gan for photo-realistic super-resolution

    JoonKyu Park, Sanghyun Son, and Kyoung Mu Lee. Content-aware local gan for photo-realistic super-resolution. In ICCV, 2023. 1

    2023

  22. [30]

    Spire: Semantic prompt-driven image restoration

    Chenyang Qi, Zhengzhong Tu, Keren Ye, Mauricio Delbra- cio, Peyman Milanfar, Qifeng Chen, and Hossein Talebi. Spire: Semantic prompt-driven image restoration. ECCV,

  23. [31]

    Xpsr: Cross-modal priors for diffusion-based image super-resolution

    Yunpeng Qu, Kun Yuan, Kai Zhao, Qizhi Xie, Jinhua Hao, Ming Sun, and Chao Zhou. Xpsr: Cross-modal priors for diffusion-based image super-resolution. ECCV, 2024. 1, 2, 7

    2024

  24. [32]

    High-resolution image syn- thesis with latent diffusion models

    Robin Rombach, Andreas Blattmann, Dominik Lorenz, Patrick Esser, and Bj¨orn Ommer. High-resolution image syn- thesis with latent diffusion models. In CVPR, 2022. 1, 2, 5

    2022

  25. [33]

    U-net: Convolutional networks for biomedical image segmentation

    Olaf Ronneberger, Philipp Fischer, and Thomas Brox. U-net: Convolutional networks for biomedical image segmentation. In MACCAI, 2015. 5 9

    2015

  26. [34]

    Image super- resolution via iterative refinement

    Chitwan Saharia, Jonathan Ho, William Chan, Tim Sali- mans, David J Fleet, and Mohammad Norouzi. Image super- resolution via iterative refinement. TPAMI, 2022. 2

    2022

  27. [35]

    Denoising diffusion implicit models

    Jiaming Song, Chenlin Meng, and Stefano Ermon. Denoising diffusion implicit models. arXiv preprint arXiv:2010.02502, 2020. 2

    2010 arXiv

  28. [36]

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

    Radu Timofte, Eirikur Agustsson, Luc Van Gool, Ming- Hsuan Yang, and Lei Zhang. Ntire 2017 challenge on single image super-resolution: Methods and results. In CVPRW,

    2017

  29. [37]

    Sam- diffsr: Structure-modulated diffusion model for image super- resolution

    Chengcheng Wang, Zhiwei Hao, Yehui Tang, Jianyuan Guo, Yujie Yang, Kai Han, and Yunhe Wang. Sam- diffsr: Structure-modulated diffusion model for image super- resolution. arXiv preprint arXiv:2402.17133, 2024. 3

    2024 arXiv

  30. [38]

    Ex- ploring clip for assessing the look and feel of images

    Jianyi Wang, Kelvin CK Chan, and Chen Change Loy. Ex- ploring clip for assessing the look and feel of images. In AAAI, 2023. 7

    2023

  31. [39]

    Exploiting diffusion prior for real-world image super-resolution

    Jianyi Wang, Zongsheng Yue, Shangchen Zhou, Kelvin CK Chan, and Chen Change Loy. Exploiting diffusion prior for real-world image super-resolution. IJCV, 2024. 1, 2, 6, 7

    2024

  32. [40]

    Recovering realistic texture in image super-resolution by deep spatial feature transform

    Xintao Wang, Ke Yu, Chao Dong, and Chen Change Loy. Recovering realistic texture in image super-resolution by deep spatial feature transform. In CVPR, 2018. 2, 6

    2018

  33. [41]

    Real-esrgan: Training real-world blind super-resolution with pure synthetic data

    Xintao Wang, Liangbin Xie, Chao Dong, and Ying Shan. Real-esrgan: Training real-world blind super-resolution with pure synthetic data. In ICCVW, 2021. 1, 2, 6, 7

    2021

  34. [42]

    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. TIP, 2004. 6

    2004

  35. [43]

    Component divide-and-conquer for real-world image super-resolution

    Pengxu Wei, Ziwei Xie, Hannan Lu, Zongyuan Zhan, Qix- iang Ye, Wangmeng Zuo, and Liang Lin. Component divide-and-conquer for real-world image super-resolution. In ECCV, 2020. 6

    2020

  36. [44]

    Seesr: Towards semantics-aware real-world image super-resolution

    Rongyuan Wu, Tao Yang, Lingchen Sun, Zhengqiang Zhang, Shuai Li, and Lei Zhang. Seesr: Towards semantics-aware real-world image super-resolution. In CVPR, 2024. 1, 2, 3, 6, 7

    2024

  37. [45]

    Segformer: Simple and ef- ficient design for semantic segmentation with transformers

    Enze Xie, Wenhai Wang, Zhiding Yu, Anima Anandkumar, Jose M Alvarez, and Ping Luo. Segformer: Simple and ef- ficient design for semantic segmentation with transformers. NIPS, 2021. 2, 4

    2021

  38. [46]

    Maniqa: Multi-dimension attention network for no-reference image quality assessment

    Sidi Yang, Tianhe Wu, Shuwei Shi, Shanshan Lao, Yuan Gong, Mingdeng Cao, Jiahao Wang, and Yujiu Yang. Maniqa: Multi-dimension attention network for no-reference image quality assessment. In CVPR, 2022. 6

    2022

  39. [47]

    Pixel-aware stable diffusion for realistic image super-resolution and personalized stylization

    Tao Yang, Rongyuan Wu, Peiran Ren, Xuansong Xie, and Lei Zhang. Pixel-aware stable diffusion for realistic image super-resolution and personalized stylization. ECCV, 2024. 1, 2, 3, 6, 7

    2024

  40. [48]

    Scaling up to excellence: Practicing model scaling for photo- realistic image restoration in the wild

    Fanghua Yu, Jinjin Gu, Zheyuan Li, Jinfan Hu, Xiangtao Kong, Xintao Wang, Jingwen He, Yu Qiao, and Chao Dong. Scaling up to excellence: Practicing model scaling for photo- realistic image restoration in the wild. In CVPR, 2024. 1, 2, 7

    2024

  41. [49]

    Resshift: Efficient diffusion model for image super- resolution by residual shifting

    Zongsheng Yue, Jianyi Wang, and Chen Change Loy. Resshift: Efficient diffusion model for image super- resolution by residual shifting. NIPS, 2024. 2, 6, 7

    2024

  42. [50]

    Designing a practical degradation model for deep blind image super-resolution

    Kai Zhang, Jingyun Liang, Luc Van Gool, and Radu Timo- fte. Designing a practical degradation model for deep blind image super-resolution. In ICCV, 2021. 1, 2, 6, 7

    2021

  43. [51]

    Adding conditional control to text-to-image diffusion models

    Lvmin Zhang, Anyi Rao, and Maneesh Agrawala. Adding conditional control to text-to-image diffusion models. In ICCV, 2023. 2

    2023

  44. [52]

    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 CVPR, 2018. 6

    2018

  45. [53]

    Efficient long-range attention network for image super- resolution

    Xindong Zhang, Hui Zeng, Shi Guo, and Lei Zhang. Efficient long-range attention network for image super- resolution. In ECCV, 2022. 1

    2022

  46. [54]

    Recognize anything: A strong image tagging model

    Youcai Zhang, Xinyu Huang, Jinyu Ma, Zhaoyang Li, Zhaochuan Luo, Yanchun Xie, Yuzhuo Qin, Tong Luo, Yaqian Li, Shilong Liu, et al. Recognize anything: A strong image tagging model. In CVPR, 2024. 1, 2

    2024

  47. [55]

    Generalized decoding for pixel, image, and language

    Xueyan Zou, Zi-Yi Dou, Jianwei Yang, Zhe Gan, Linjie Li, Chunyuan Li, Xiyang Dai, Harkirat Behl, Jianfeng Wang, Lu Yuan, et al. Generalized decoding for pixel, image, and language. In CVPR, 2023. 2, 6, 8

    2023

  48. [56]

    Segment everything everywhere all at once

    Xueyan Zou, Jianwei Yang, Hao Zhang, Feng Li, Linjie Li, Jianfeng Wang, Lijuan Wang, Jianfeng Gao, and Yong Jae Lee. Segment everything everywhere all at once. NIPS, 36,

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.