To play video, one should be considered: compatibility between media devices (smart phone, TV, black box, etc.). And we need to specify which format should be used to deliver/store the media using cutting-edge technologies. For that, the international standardization groups, MPEG and VCEG, lead several meetings per year.

We do research on the recent video coding standardization projects including HEVC, VVC, MPEG-I: Immersive Video, and also on deep learning-related projects like VCM (Video coding for machine). Not only conducting research for that, but also present ours to the meetings to adapt ours as international standard. In addition, we investigate which direction of video coding standard is for future industry.

Now deep learning research scope is enlarged from simple image to video contents, and its model is going deeper and wider. Whereas this model can contribute on increasing the accuracy of image classification/interpretation/prediction, it needs expensive and huge devices, thereby having many difficulties to use the model in real world. Thus, less complex model is needed for smart phone, vehicle, drone, and IoT devices where battery is constrained, though these devices can be reached easily at people. Accordingly, we investigate the possibility to compress the model and to speed-up the model.