Title: Compression artifacts removal using convolutional neural networks
Authors: Svoboda, Pavel
Hradiš, Michal
Bařina, David
Zemčík, Pavel
Citation: Journal of WSCG. 2016, vol. 24, no. 2, p. 63-72.
Issue Date: 2016
Publisher: Václav Skala - UNION Agency
Document type: článek
URI: http://wscg.zcu.cz/WSCG2016/!_2016_Journal_WSCG-No-2.pdf
ISSN: 1213-6972 (print)
1213-6980 (CD-ROM)
1213-6964 (on-line)
Keywords: hluboké učení;konvoluční neuronová síť;JPEG
Keywords in different language: deep learning;convolutional neural networks;JPEG
Abstract in different language: This paper shows that it is possible to train large and deep convolutional neural networks (CNN) for JPEG compression artifacts reduction, and that such networks can provide significantly better reconstruction quality compared to previously used smaller networks as well as to any other state-of-the-art methods. We were able to train networks with 8 layers in a single step and in relatively short time by combining residual learning, skip architecture, and symmetric weight initialization. We provide further insights into convolution networks for JPEG artifact reduction by evaluating three different objectives, generalization with respect to training dataset size, and generalization with respect to JPEG quality level.
Rights: © Václav Skala - UNION Agency
Appears in Collections:Volume 24, Number 2 (2016)

Files in This Item:
File Description SizeFormat 
Svoboda.pdfPlný text2,67 MBAdobe PDFView/Open

Please use this identifier to cite or link to this item: http://hdl.handle.net/11025/21649

Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.