mlsanityData to 14 Oct 2026

OracleZoom: On-Policy Self-Distillation Inspired Reference-Constrained Recursive Image Super Resolution

arXiv 2609.06490 ↗First posted 6 Sep 2026cs.CV cs.AI cs.CLcited by 1

On the map: Knowledge distillation → General knowledge distillation methods

Abstract (arXiv)

Recursive Super-Resolution (SR) extends fixed-scale SR to extreme magnification by repeatedly feeding predictions back into the same model, analogous to zooming an image repeatedly. However, ground truth availability at every scale, especially at depth, remains challenging as the required source resolution grows geometrically, leaving deeper predictions unsupervised. We present OracleZoom, an on-policy distillation-inspired, reference-constrained framework that trains on its trajectory while carrying the last ground-truth evidence beyond the supervision boundary. Direct and cross-scale supervision constrain verifiable content, while a no-reference quality objective guides unresolved fine-scale detail. A KL-constrained pretrained latent prior limits quality-driven drift, while EMA consistency stabilizes the supervision boundary. Across seven datasets, OracleZoom achieves the state-of-the-art SR quality across zooming scales, averaging 0.713 CLIPIQA, with larger gains on deeper scales, while significantly reducing hallucinations. Code, data, and models are available at https://dipta007.github.io/OracleZoom/ .

Key facts. Cited by 1 paper in the map (data through 14 Oct 2026). Placed in General knowledge distillation methods, one of 12 directions in Knowledge distillation. Builds on 75 of its 93 references that are in the map. First posted on arXiv 6 Sep 2026.

Builds on

75 of 93 references.

References · 75

  1. 2021LoRA: Low-Rank Adaptation of Large Language Modelscited by 14,380
  2. 2018The Unreasonable Effectiveness of Deep Features as a Perceptual Metriccited by 11,177
  3. 2015Distilling the Knowledge in a Neural Networkcited by 10,324
  4. 2021Diffusion Models Beat GANs on Image Synthesiscited by 8,255
  5. 2023DINOv2: Learning Robust Visual Features without Supervisioncited by 7,974
  6. 2021Emerging Properties in Self-Supervised Vision Transformerscited by 6,968
  7. 2018A Style-Based Generator Architecture for Generative Adversarial Networkscited by 6,013
  8. 2021SwinIR: Image Restoration Using Swin Transformercited by 1,887
  9. 2017Enhanced Deep Residual Networks for Single Image Super-Resolutioncited by 1,552
  10. 2025InternVL3.5: Advancing Open-Source Multimodal Models in Versatility, Reasoning, and Efficiencycited by 1,510
  11. 2019Self-training with Noisy Student improves ImageNet classificationcited by 1,311
  12. 2018ESRGAN: Enhanced Super-Resolution Generative Adversarial Networkscited by 1,254
38 more
  1. 2021MUSIQ: Multi-scale Image Quality Transformercited by 1,081
  2. 2022Exploring CLIP for Assessing the Look and Feel of Imagescited by 991
  3. 2021Real-ESRGAN: Training Real-World Blind Super-Resolution with Pure Synthetic Datacited by 970
  4. 2020Image Quality Assessment: Unifying Structure and Texture Similaritycited by 811
  5. 2015Scheduled Sampling for Sequence Prediction with Recurrent Neural Networkscited by 803
  6. 2016Sequence-Level Knowledge Distillationcited by 801
  7. 2017The Perception-Distortion Tradeoffcited by 587
  8. 2020Learning Continuous Image Representation with Local Implicit Image Functioncited by 552
  9. 2021Designing a Practical Degradation Model for Deep Blind Image Super-Resolutioncited by 529
  10. 2022MANIQA: Multi-dimension Attention Network for No-Reference Image Quality Assessmentcited by 522
  11. 2023Exploiting Diffusion Prior for Real-World Image Super-Resolutioncited by 489
  12. 2017Deep Laplacian Pyramid Networks for Fast and Accurate Super-Resolutioncited by 453
  13. 2015Policy Distillationcited by 440
  14. 2019Toward Real-World Single Image Super-Resolution: A New Benchmark and A New Modelcited by 403
  15. 2023ResShift: Efficient Diffusion Model for Image Super-resolution by Residual Shiftingcited by 365
  16. 2023Uni-ControlNet: All-in-One Control to Text-to-Image Diffusion Modelscited by 359
  17. 2024MambaIR: A Simple Baseline for Image Restoration with State-Space Modelcited by 352
  18. 2023SeeSR: Towards Semantics-Aware Real-World Image Super-Resolutioncited by 315
  19. 2024Scaling Up to Excellence: Practicing Model Scaling for Photo-Realistic Image Restoration In the Wildcited by 286
  20. 2023Pixel-Aware Stable Diffusion for Realistic Image Super-resolution and Personalized Stylizationcited by 254
  21. 2020Component Divide-and-Conquer for Real-World Image Super-Resolutioncited by 246
  22. 2024One-Step Effective Diffusion Network for Real-World Image Super-Resolutioncited by 246
  23. 2023SinSR: Diffusion-Based Image Super-Resolution in a Single Stepcited by 238
  24. 2015Unifying distillation and privileged informationcited by 233
  25. 2023TOPIQ: A Top-down Approach from Semantics to Distortions for Image Quality Assessmentcited by 226
  26. 2023Implicit Diffusion Models for Continuous Super-Resolutioncited by 218
  27. 2016Professor Forcing: A New Algorithm for Training Recurrent Networkscited by 216
  28. 2023Generative Diffusion Prior for Unified Image Restoration and Enhancementcited by 199
  29. 2019Meta-SR: A Magnification-Arbitrary Network for Super-Resolutioncited by 181
  30. 2021Unsupervised Degradation Representation Learning for Blind Super-Resolutioncited by 177
  31. 2021Local Texture Estimator for Implicit Representation Functioncited by 127
  32. 2024Controllable Generation with Text-to-Image Diffusion Models: A Surveycited by 83
  33. 2026Vision-OPD: Learning to See Fine Details for Multimodal LLMs via On-Policy Self-Distillationcited by 77
  34. 2022CiaoSR: Continuous Implicit Attention-in-Attention Network for Arbitrary-Scale Image Super-Resolutioncited by 70
  35. 2023CoSeR: Bridging Image and Language for Cognitive Super-Resolutioncited by 66
  36. 2023Cascaded Local Implicit Transformer for Arbitrary-Scale Super-Resolutioncited by 52
  37. 2024HiT-SR: Hierarchical Transformer for Efficient Image Super-Resolutioncited by 42
  38. 2025VOLD: Reasoning Transfer from LLMs to Vision-Language Models via On-Policy Distillationcited by 31

50 most cited of 75 All 75 (JSON)

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2026: 1.

1 paper.

  1. 2026New Evidence, Same Choice: Testing Physical Experiment Selection in Vision Language Modelscited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2024You Only Need One Step: Fast Super-Resolution with Stable Diffusion via Scale Distillation
  2. 2026PhoenixSR: Generative Heterogeneous Distillation Unleashes Efficient Models for Real-World Super-Resolution
  3. 2025CTSR: Controllable Fidelity-Realness Trade-off Distillation for Real-World Image Super Resolution
  4. 2023Dissecting Arbitrary-scale Super-resolution Capability from Pre-trained Diffusion Generative Models
  5. 2025OmniScaleSR: Unleashing Scale-Controlled Diffusion Prior for Faithful and Realistic Arbitrary-Scale Image Super-Resolution
  6. 2025Local-Global Context-Aware and Structure-Preserving Image Super-Resolution
  7. 2025Iterative Inference-time Scaling with Adaptive Frequency Steering for Image Super-Resolution
  8. 2024ControlSR: Taming Diffusion Models for Consistent Real-World Image Super Resolution