| Paper Abstract and Keywords |
| Presentation |
2020-02-27 13:00
Video Coding Using Optimal Intra Prediction Mode Estimation by CNN
Ryota Yokoyama, Masahiko Tahara (Waseda Univ.), Heming Sun (Waseda Univ./JST), Masaru Takeuchi (Waseda Univ.), Yasutaka Matsuo (NHK), Jiro Katto (Waseda Univ.) |
| Abstract |
(in Japanese) |
(See Japanese page) |
| (in English) |
These days, efficient video coding is required due to spread of video production and viewing, and high definition video. In order to select the Rate-Distortion (RD) optimized intra prediction mode, RD cost is used in High Efficiency Video Coding Test Model (HM). However, since calculation of RD cost is computationally intensive, intra prediction modes are filtered to the candidates for RD cost calculation by alternative cost calculation or Most Probable Mode (MPM) with low computational complexity. We introduce video coding using MPMs obtained by Convolutional Neural Network (CNN). As experimental results, we confirmed that our method achieve higher accuracy in optimal intra prediction mode estimation and 0.16% reduction in Y BD-Rate, compared to HM. |
| Keyword |
(in Japanese) |
(See Japanese page) |
| (in English) |
Intra prediction / Convolutional Neural Network (CNN) / Most Probable Mode (MPM) / Video coding / / / / |
| Reference Info. |
ITE Tech. Rep. |
| Paper # |
|
| Date of Issue |
|
| ISSN |
|
| Download PDF |
|