https://scholars.lib.ntu.edu.tw/handle/123456789/607244
標題: | How to exploit the transferability of learned image compression to conventional codecs | 作者: | Klopp J.P Liu K.-C LIANG-GEE CHEN SHAO-YI CHIEN |
關鍵字: | Convolutional neural networks;Encoding (symbols);Image coding;Image compression;Textures;Convolutional neural network;Decoding overheads;Images compression;Lossy image compression;Modal loss;Multi-modal;Network-based algorithm;Performance measure;Recent researches;Task-aware;Generative adversarial networks | 公開日期: | 2021 | 起(迄)頁: | 16160-16169 | 來源出版物: | Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition | 會議論文: | 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition, CVPR 2021 | 摘要: | Lossy image compression is often limited by the simplicity of the chosen loss measure. Recent research suggests that generative adversarial networks have the ability to overcome this limitation and serve as a multi-modal loss, especially for textures. Together with learned image compression, these two techniques can be used to great effect when relaxing the commonly employed tight measures of distortion. However, convolutional neural network-based algorithms have a large computational footprint. Ideally, an existing conventional codec should stay in place, ensuring faster adoption and adherence to a balanced computational envelope. As a possible avenue to this goal, we propose and investigate how learned image coding can be used as a surrogate to optimise an image for encoding. A learned filter alters the image to optimise a different performance measure or a particular task. Extending this idea with a generative adversarial network, we show how entire textures are replaced by ones that are less costly to encode but preserve a sense of detail. Our approach can remodel a conventional codec to adjust for the MS-SSIM distortion with over 20% rate improvement without any decoding overhead. On task-aware image compression, we perform favourably against a similar but codec-specific approach. ? 2021 IEEE |
URI: | https://www.scopus.com/inward/record.uri?eid=2-s2.0-85121803469&doi=10.1109%2fCVPR46437.2021.01590&partnerID=40&md5=d1688f0d5e830c8287d356d605e31c9b https://scholars.lib.ntu.edu.tw/handle/123456789/607244 |
ISSN: | 10636919 | DOI: | 10.1109/CVPR46437.2021.01590 |
顯示於: | 電機工程學系 |
在 IR 系統中的文件,除了特別指名其著作權條款之外,均受到著作權保護,並且保留所有的權利。