Search by item HOME > Access full text > Search by item

JBE, vol. 23, no. 5, pp.718-721, September, 2018

DOI: https://doi.org/10.5909/JBE.2018.23.5.718

Deep Learning based Inter Prediction Technique for Video Coding

Jeongkyung Lee, Nayoung Kim, and Je-Won Kang

C.A E-mail: jewonk@ewha.ac.kr

Abstract:

This paper presents an inter-prediction technique using deep learning, where a virtual reference frame of the current frame is synthesized by using the reconstructed frames to improve coding efficiency. Experimental results demonstrate that the proposed algorithm provides 1.9% BD-rate reduction on average as compared to HEVC reference software in the Random Access condition.

 



Keyword: HEVC, deep learning, virtual reference frame

Reference:
[1] S. Liu, L. Wang, P. Wu, and H. Yang, “JVET AHG report 9: Neural Networks in Video Coding (AHG9)” in ISO/IEC/JTC1/SC29/ WG11 and ITU-T SG16 Q.6, Apr. 2018.
[2] S.H Cho, Y. H. Kim, W. L, H.W. Kim, and C.S. Choi, “A Technical Analysis on Deep Learning based Image and Video Compression", Journal of Broadcast Engineering, Vol. 23, No. 3, 383-394, May 2018
[3] W. Park and M. Kim, "CNN-based in-loop filtering for coding efficiency improvement," IEEE Image, Video, and Multidimensional Signal Processing Workshop, 2016.
[4] J. Kang, S. Kim, and K. M. Lee, “Multi-modal Multi-scale Convolu- tional Neural Network based In-loop Filter Design for Next Generation Video Codec,” IEEE International Conference on Image Processing, 2017.
[5] J.K. Lee and J.-W. Kang, “Video coding technique based on deep learning”, 2018 KIBME Summer Conference.
[6] B. D. Brabandere, X. Jia, T. Tuytelaars, and L. V. Gool, "Dynamic filter networks," Neural Information Processing Systems (NIPS). 2016.
[7] S. Niklaus, L. Mai, and F. Liu. "Video frame interpolation via adaptive separable convolution," International Conference on Computer Vision, 2017.
[8] HM16.9 software, available at: https://hevc.hhi.fraunhofer.de/svn /svn_HEVCSoftware/tags/HM-16.9
[9] F. Bossen, “JCTVC-L1100: Common test conditions and software reference configurations” in ISO/IEC/JTC1/SC29/WG11 and ITU-T SG16 Q.6, Jan.2013.

Comment


Editorial Office
1108, New building, 22, Teheran-ro 7-gil, Gangnam-gu, Seoul, Korea
Homepage: www.kibme.org TEL: +82-2-568-3556 FAX: +82-2-568-3557
Copyrightⓒ 2012 The Korean Institute of Broadcast and Media Engineers
All Rights Reserved