PulseAugur
LIVE 03:38:35
research · [9 sources] ·
0
research

New diffusion models tackle image super-resolution with wavelet and latent space innovations

By PulseAugur Editorial · Summary by gemini-2.5-flash-lite from 9 sources

Researchers have developed two new frameworks, SlimDiffSR and TOC-SR, to make diffusion models more efficient for image super-resolution tasks. SlimDiffSR focuses on remote sensing imagery by using a distilled teacher model and structured pruning techniques, achieving up to a 200x inference acceleration and a 20x reduction in parameters. TOC-SR creates compact diffusion backbones through feature-wise distillation and architecture discovery, resulting in a 6.6x parameter reduction and a 2.8x reduction in GMACs before distilling into a single-step generator. Both approaches aim to balance high reconstruction quality with significantly reduced computational costs for practical deployment. AI

Summary written by gemini-2.5-flash-lite from 9 sources. How we write summaries →

IMPACT These advancements could enable wider adoption of diffusion models for image enhancement tasks by reducing computational requirements.

RANK_REASON Two new research papers introduce methods to make diffusion models more efficient for image super-resolution.

Read on arXiv cs.CV →

COVERAGE [9]

  1. arXiv cs.LG TIER_1 · Onkar Jadhav, Tim French, Matthew Rayson, Nicole L. Jones ·

    PODiff: Latent Diffusion in Proper Orthogonal Decomposition Space for Scientific Super-Resolution

    arXiv:2605.03399v1 Announce Type: new Abstract: Probabilistic super-resolution of high-dimensional spatial fields using diffusion models is often computationally prohibitive due to the cost of operating directly in pixel space. We propose PODiff, a structured conditional generati…

  2. arXiv cs.CV TIER_1 · Lorenzo Aloisi, Luigi Sigillo, Aurelio Uncini, Danilo Comminiello ·

    A Wavelet Diffusion GAN for Image Super-Resolution

    arXiv:2410.17966v3 Announce Type: replace-cross Abstract: In recent years, diffusion models have emerged as a superior alternative to generative adversarial networks (GANs) for high-fidelity image generation, with wide applications in text-to-image generation, image-to-image tran…

  3. arXiv cs.CV TIER_1 · Luigi Sigillo, Christian Bianchi, Aurelio Uncini, Danilo Comminiello ·

    Quaternion Wavelet-Conditioned Diffusion Models for Image Super-Resolution

    arXiv:2505.00334v3 Announce Type: replace Abstract: Image Super-Resolution is a fundamental problem in computer vision with broad applications spacing from medical imaging to satellite analysis. The ability to reconstruct high-resolution images from low-resolution inputs is cruci…

  4. arXiv cs.CV TIER_1 · Luigi Sigillo, Renato Giamba, Danilo Comminiello ·

    Metadata, Wavelet, and Time Aware Diffusion Models for Satellite Image Super Resolution

    arXiv:2506.23566v2 Announce Type: replace Abstract: The acquisition of high-resolution satellite imagery is often constrained by the spatial and temporal limitations of satellite sensors, as well as the high costs associated with frequent observations. These challenges hinder app…

  5. arXiv cs.CV TIER_1 · Tayyab Nasir, Daochang Liu, Ajmal Mian ·

    Implicit Neural Representation-Based Continuous Single Image Super-Resolution: An Empirical Benchmark

    arXiv:2601.17723v2 Announce Type: replace Abstract: Implicit neural representation (INR) has become the standard approach for arbitrary-scale image super-resolution (ASSR). To date, no empirical study has systematically examined the effectiveness of existing methods, nor investig…

  6. arXiv cs.CV TIER_1 · Ce Wang, Zhenyu Hu, Wanjie Sun ·

    SlimDiffSR: Toward Lightweight and Efficient Remote Sensing Image Super-Resolution via Diffusion Model Distillation

    arXiv:2605.02198v1 Announce Type: new Abstract: Diffusion models have recently achieved remarkable performance in image super-resolution (SR), but their high computational cost limits practical deployment in remote sensing applications. To address this issue, we propose SlimDiffS…

  7. arXiv cs.CV TIER_1 · Sowmya Vajrala, Akshay Bankar, Manjunath Arveti, Shreyas Pandith, Sravanth Kodavanti, Subhajit Sanyal, Amit Unde, Srinivas Soumitri Miriyala ·

    TOC-SR: Task-Optimal Compact diffusion for Image Super Resolution

    arXiv:2605.02767v1 Announce Type: new Abstract: Diffusion models have recently demonstrated strong performance for image restoration tasks, including super-resolution. However, their large model size and iterative sampling procedures make them computationally expensive for practi…

  8. arXiv cs.CV TIER_1 · Srinivas Soumitri Miriyala ·

    TOC-SR: Task-Optimal Compact diffusion for Image Super Resolution

    Diffusion models have recently demonstrated strong performance for image restoration tasks, including super-resolution. However, their large model size and iterative sampling procedures make them computationally expensive for practical deployment. In this work, we present TOC-SR,…

  9. arXiv cs.CV TIER_1 · Wanjie Sun ·

    SlimDiffSR: Toward Lightweight and Efficient Remote Sensing Image Super-Resolution via Diffusion Model Distillation

    Diffusion models have recently achieved remarkable performance in image super-resolution (SR), but their high computational cost limits practical deployment in remote sensing applications. To address this issue, we propose SlimDiffSR, a lightweight and efficient diffusion-based f…

RELATED ENTITIES

RELATED TOPICS