{"title":"R-ACE Network for OLED Image Power Saving","authors":"Kuntoro Adi Nugroho, S. Ruan","doi":"10.1109/LifeTech53646.2022.9754748","DOIUrl":null,"url":null,"abstract":"Energy efficient display is essential to significantly reduce the power consumption of a device. OLED is an emissive display in which power consumption highly depends on pixel intensities of the emitted image. We propose two improvements for ACE network to produce low power image. First, we utilize L2 power-loss which results in faster training. Secondly, we add residual connection to the network and process the original luminance image. Evaluated in LIVE and BSD dataset with power reduction rate of 0.4, 0.6, and 0.8, our proposed method outperformed the original ACE network in most scenario.","PeriodicalId":297484,"journal":{"name":"2022 IEEE 4th Global Conference on Life Sciences and Technologies (LifeTech)","volume":"223 1","pages":"0"},"PeriodicalIF":0.0000,"publicationDate":"2022-03-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"5","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"2022 IEEE 4th Global Conference on Life Sciences and Technologies (LifeTech)","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1109/LifeTech53646.2022.9754748","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
引用次数: 5
Abstract
Energy efficient display is essential to significantly reduce the power consumption of a device. OLED is an emissive display in which power consumption highly depends on pixel intensities of the emitted image. We propose two improvements for ACE network to produce low power image. First, we utilize L2 power-loss which results in faster training. Secondly, we add residual connection to the network and process the original luminance image. Evaluated in LIVE and BSD dataset with power reduction rate of 0.4, 0.6, and 0.8, our proposed method outperformed the original ACE network in most scenario.