OracleZoom: On-Policy Self-Distillation Inspired Reference-Constrained Recursive Image Super Resolution
OracleZoom enables recursive extreme-scale image super-resolution via reference-constrained on-policy distillation, reducing hallucinations at deep zoom scales.
OracleZoom tackles recursive super-resolution, where repeated feeding of predictions back into the same model leaves deeper-scale outputs unsupervised as required source resolution grows geometrically. The framework trains on its own trajectory while carrying the last ground-truth evidence beyond the supervision boundary, combining direct and cross-scale supervision, a no-reference quality objective, a KL-constrained pretrained latent prior, and EMA consistency. Across seven datasets it achieves state-of-the-art SR quality across zoom scales, averaging 0.713 CLIPIQA with larger gains at deeper scales and significantly reduced hallucinations. Code, data, and models are publicly released.
- Addresses unsupervised deeper scales in recursive image super-resolution
- Combines reference-constrained supervision with no-reference quality guidance
- KL-constrained latent prior limits quality-driven drift; EMA consistency stabilizes training
- State-of-the-art across seven datasets averaging 0.713 CLIPIQA
- Code, data, and models publicly available
From the original article
1 min read at huggingface.co
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.
By Shubhashis Roy Dipta, Sourajit Saha, Shaswati Saha · huggingface.co
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.