mlsanityData to 7 Oct 2026

UniRestorer: Universal Image Restoration via Adaptively Estimating Image Degradation at Proper Granularity

arXiv 2412.20157 ↗First posted 28 Dec 2024cs.CVcited by 12

On the map: Mixture-of-experts → Vision mixture-of-experts

Abstract (arXiv)

Recently, considerable progress has been made in all-in-one image restoration. Generally, existing methods can be degradation-agnostic or degradation-aware. However, the former are limited in leveraging degradation-specific restoration, and the latter suffer from the inevitable error in degradation estimation. Consequently, the performance of existing methods has a large gap compared to specific single-task models. In this work, we make a step forward in this topic, and present our UniRestorer with improved restoration performance. Specifically, we perform hierarchical clustering on degradation space, and train a multi-granularity mixture-of-experts (MoE) restoration model. Then, UniRestorer adopts both degradation and granularity estimation to adaptively select an appropriate expert for image restoration. In contrast to existing degradation-agnostic and -aware methods, UniRestorer can leverage degradation estimation to benefit degradation specific restoration, and use granularity estimation to make the model robust to degradation estimation error. Experimental results show that our UniRestorer outperforms state-of-the-art all-in-one methods by a large margin, and is promising in closing the performance gap to specific single task models.

Key facts. Cited by 12 papers in the map (data through 7 Oct 2026). Placed in Vision mixture-of-experts, one of 12 directions in Mixture-of-experts. Builds on 83 of its 114 references that are in the map. First posted on arXiv 28 Dec 2024.

Builds on

83 of 114 references.

Closest ancestors · 7

Same direction, newest first

  1. 2024Complexity Experts are Task-Discriminative Learners for Any Image Restoration27 Nov
  2. 2024LoRA-IR: Taming Low-Rank Experts for Efficient All-in-One Image Restoration20 Oct
  3. 2023From Sparse to Soft Mixtures of Experts2 Aug
  4. 2023An Efficient General-Purpose Modular Vision Model via Multi-Task Heterogeneous Training29 Jun
  5. 2023Scaling Vision-Language Models with Sparse Mixture of Experts13 Mar
  6. 2022Residual Mixture of Experts20 Apr
  7. 2021Scaling Vision with Sparse Mixture of Experts10 Jun

Other references · 76

  1. 2017Attention Is All You Needcited by 59,932
  2. 2015Deep Residual Learning for Image Recognitioncited by 51,809
  3. 2021Learning Transferable Visual Models From Natural Language Supervisioncited by 33,456
  4. 2014Very Deep Convolutional Networks for Large-Scale Image Recognitioncited by 15,267
  5. 2021LoRA: Low-Rank Adaptation of Large Language Modelscited by 14,084
  6. 2021Swin Transformer: Hierarchical Vision Transformer using Shifted Windowscited by 12,082
  7. 2023BLIP-2: Bootstrapping Language-Image Pre-training with Frozen Image Encoders and Large Language Modelscited by 6,069
  8. 2022BLIP: Bootstrapping Language-Image Pre-training for Unified Vision-Language Understanding and Generationcited by 4,310
  9. 2019Parameter-Efficient Transfer Learning for NLPcited by 4,195
  10. 2021The Power of Scale for Parameter-Efficient Prompt Tuningcited by 3,769
  11. 2017Outrageously Large Neural Networks: The Sparsely-Gated Mixture-of-Experts Layercited by 3,413
  12. 2021Switch Transformers: Scaling to Trillion Parameter Models with Simple and Efficient Sparsitycited by 3,014
38 more
  1. 2021SwinIR: Image Restoration Using Swin Transformercited by 1,869
  2. 2021Restormer: Efficient Transformer for High-Resolution Image Restorationcited by 1,618
  3. 2020GShard: Scaling Giant Models with Conditional Computation and Automatic Shardingcited by 1,601
  4. 2017Enhanced Deep Residual Networks for Single Image Super-Resolutioncited by 1,543
  5. 2016Beyond a Gaussian Denoiser: Residual Learning of Deep CNN for Image Denoisingcited by 1,352
  6. 2021Real-ESRGAN: Training Real-World Blind Super-Resolution with Pure Synthetic Datacited by 961
  7. 2021Multi-Stage Progressive Image Restorationcited by 813
  8. 2016Deep Multi-scale Convolutional Neural Network for Dynamic Scene Deblurringcited by 722
  9. 2018Deep Retinex Decomposition for Low-Light Enhancementcited by 720
  10. 2022Simple Baselines for Image Restorationcited by 701
  11. 2021Designing a Practical Degradation Model for Deep Blind Image Super-Resolutioncited by 524
  12. 2017Benchmarking Single Image Dehazing and Beyondcited by 445
  13. 2022DeepSpeed-MoE: Advancing Mixture-of-Experts Inference and Training to Power Next-Generation AI Scalecited by 400
  14. 2022ST-MoE: Designing Stable and Transferable Sparse Expert Modelscited by 391
  15. 2019DeblurGAN-v2: Deblurring (Orders-of-Magnitude) Faster and Bettercited by 361
  16. 2016Deep Joint Rain Detection and Removal from a Single Imagecited by 351
  17. 2023ControlVideo: Training-free Controllable Text-to-Video Generationcited by 324
  18. 2021BASE Layers: Simplifying Training of Large, Sparse Modelscited by 315
  19. 2018Additive Margin Softmax for Face Verificationcited by 275
  20. 2019Progressive Image Deraining Networks: A Better and Simpler Baselinecited by 271
  21. 2022Vision Transformers for Single Image Dehazingcited by 255
  22. 2019An Underwater Image Enhancement Benchmark Dataset and Beyondcited by 250
  23. 2022Unified Scaling Laws for Routed Language Modelscited by 249
  24. 2018Density-aware Single Image De-raining using a Multi-stream Dense Networkcited by 248
  25. 2021TransWeather: Transformer-based Restoration of Images Degraded by Adverse Weather Conditionscited by 243
  26. 2023Retinexformer: One-stage Retinex-based Transformer for Low-light Image Enhancementcited by 228
  27. 2017Learning a Single Convolutional Super-Resolution Network for Multiple Degradationscited by 227
  28. 2017Attentive Generative Adversarial Network for Raindrop Removal from a Single Imagecited by 204
  29. 2022Multimodal Contrastive Learning with LIMoE: the Language-Image Mixture of Expertscited by 202
  30. 2017DesnowNet: Context-Aware Deep Network for Snow Removalcited by 200
  31. 2024Mixture of LoRA Expertscited by 179
  32. 2024InstructIR: High-Quality Image Restoration Following Human Instructionscited by 139
  33. 2020Data Uncertainty Learning in Face Recognitioncited by 134
  34. 2022Efficient and Degradation-Adaptive Network for Real-World Image Super-Resolutioncited by 96
  35. 2024OneRestore: A Universal Restoration Framework for Composite Degradationcited by 91
  36. 2019A General Decoupled Learning Framework for Parameterized Image Operatorscited by 72
  37. 2024Convolution Meets LoRA: Parameter Efficient Finetuning for Segment Anything Modelcited by 67
  38. 2022Uni-Perceiver-MoE: Learning Sparse Generalist Models with Conditional MoEscited by 58

50 most cited of 76 All 76 (JSON)

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2025: 5; 2026: 7.

12 papers.

  1. 2025M2Restore: Mixture-of-Experts-based Mamba-CNN Fusion Framework for All-in-One Image Restorationcited by 6
  2. 2025ClusIR: Towards Cluster-Guided All-in-One Image Restorationcited by 4
  3. 2025Towards Unified Image Deblurring using a Mixture-of-Experts Decodercited by 2
  4. 2026SplatWeaver: Learning to Allocate Gaussian Primitives for Generalizable Novel View Synthesiscited by 2
  5. 2026PixRestore: Unified Image Restoration via Pixel Diffusion Transformercited by 2
  6. 2025SelfHVD: Self-Supervised Handheld Video Deblurringcited by 0
  7. 2025Unleashing Degradation-Carrying Features in Symmetric U-Net: Simpler and Stronger Baselines for All-in-One Image Restorationcited by 0
  8. 2026MiM-DiT: MoE in MoE with Diffusion Transformers for All-in-One Image Restorationcited by 0
  9. 2026Compositional-Degradation UAV Image Restoration: Conditional Decoupled MoE Network and A Benchmarkcited by 0
  10. 2026GS-RealBlur: A Flexible Data Acquisition Framework for Real-World Image Deblurringcited by 0
  11. 2026TaskIR: Task-Driven Image Restoration via Degradation Adaptation and Task Feedbackcited by 0
  12. 2026Learning from Failure: Leveraging Unreliable Predictions in Semi-Supervised Real-World Adverse Weather Removalcited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2025UniRes: Universal Image Restoration for Complex Degradations
  2. 2026UnSCAR: Universal, Scalable, Controllable, and Adaptable Image Restoration
  3. 2024Chain-of-Restoration: Multi-Task Image Restoration Models are Zero-Shot Step-by-Step Universal Image Restorers
  4. 2025UniLDiff: Unlocking the Power of Diffusion Priors for All-in-One Image Restoration
  5. 2025UniRestore: Unified Perceptual and Task-Oriented Image Restoration Model Using Diffusion Prior
  6. 2025UnfoldLDM: Degradation-Aware Unfolding with Iterative Latent Diffusion Priors for Blind Image Restoration
  7. 2025DPMambaIR: All-in-One Image Restoration via Degradation-Aware Prompt State Space Model
  8. 2026Continuous Expert Assembly: Instance-Conditioned Low-Rank Residuals for All-in-One Image Restoration