Zhiyong Li, Sangjin Kim, Dongseok Im, Donghyeon Han, H. Yoo
{"title":"An 0.92 mJ/frame High-quality FHD Super-resolution Mobile Accelerator SoC with Hybrid-precision and Energy-efficient Cache","authors":"Zhiyong Li, Sangjin Kim, Dongseok Im, Donghyeon Han, H. Yoo","doi":"10.1109/CICC53496.2022.9772778","DOIUrl":null,"url":null,"abstract":"With the rise of contactless communication and streaming services, Super-resolution (SR) in mobile devices has become one of the most important image processing technologies. Also, The popularity of high-end Application Processor (AP) and high resolution display in mobile drives the development of the lightweight mobile SR-CNNs [1], [2], which show the high reconstruction quality. However, the large size and wide dynamic range of both images and intermediate feature maps in CNN hidden layers pose challenges for mobile platforms. Constraints from the limited power $(< 100\\text{mW})$ and shared bandwidth $(< 2\\text{GB}/\\mathrm{s})$ on mobile platform, a low power and energy-efficient architecture is required.","PeriodicalId":415990,"journal":{"name":"2022 IEEE Custom Integrated Circuits Conference (CICC)","volume":"16 1","pages":"0"},"PeriodicalIF":0.0000,"publicationDate":"2022-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"2","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"2022 IEEE Custom Integrated Circuits Conference (CICC)","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1109/CICC53496.2022.9772778","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
引用次数: 2
Abstract
With the rise of contactless communication and streaming services, Super-resolution (SR) in mobile devices has become one of the most important image processing technologies. Also, The popularity of high-end Application Processor (AP) and high resolution display in mobile drives the development of the lightweight mobile SR-CNNs [1], [2], which show the high reconstruction quality. However, the large size and wide dynamic range of both images and intermediate feature maps in CNN hidden layers pose challenges for mobile platforms. Constraints from the limited power $(< 100\text{mW})$ and shared bandwidth $(< 2\text{GB}/\mathrm{s})$ on mobile platform, a low power and energy-efficient architecture is required.