CiaoSR: Continuous Implicit Attention-in-Attention Network for Arbitrary-Scale Image Super-Resolution

Cao, Jiezhang; Wang, Qin; Xian, Yongqin; Li, Yawei; Ni, Bingbing; Pi, Zhiming; Zhang, Kai; Zhang, Yulun; Timofte, Radu; Gool, Luc Van

doi:10.48550/arxiv.2212.04362

Cited by 1 publication

(1 citation statement)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…However, in real-world scenarios, scaling up LR videos with user-desired scales has more practical value. Recent work on arbitrary-scale single-image super-resolution (Hu et al 2019;Chen, Liu, and Wang 2021;Lee and Jin 2022;Cao et al 2022) has explored the possibility of replacing the pixelshuffle-based upsampling module and supporting arbitrary-scale super-resolution. These methods can be divided into two categories: the implicit neural function-based (Chen, Liu, and Wang 2021;Lee and Jin 2022;Cao et al 2022) and the filter-based (Hu et al 2019) approaches.…”

Section: Introductionmentioning

confidence: 99%

Arbitrary-Scale Video Super-resolution Guided by Dynamic Context

Huang,

Li,

Chu

et al. 2024

AAAI

View full text Add to dashboard Cite

We propose a Dynamic Context-Guided Upsampling (DCGU) module for video super-resolution (VSR) that leverages temporal context guidance to achieve efficient and effective arbitrary-scale VSR. While most VSR research focuses on backbone design, the importance of the upsampling part is often overlooked. Existing methods rely on pixelshuffle-based upsampling, which has limited capabilities in handling arbitrary upsampling scales. Recent attempts to replace pixelshuffle-based modules with implicit neural function-based and filter-based approaches suffer from slow inference speeds and limited representation capacity, respectively. To overcome these limitations, our DCGU module predicts non-local sampling locations and content-dependent filter weights, enabling efficient and effective arbitrary-scale VSR. Our proposed multi-granularity location search module efficiently identifies non-local sampling locations across the entire low-resolution grid, and the temporal bilateral filter modulation module integrates content information with the filter weight to enhance textual details. Extensive experiments demonstrate the superiority of our method in terms of performance and speed on arbitrary-scale VSR.

show abstract