Popularity Prediction for Social Media over Arbitrary Time Horizons
Daniel Haimovich, Dima Karamshuk, Thomas Leeper, Evgeniy Riabenko, Milan Vojnovic
International Conference on Artificial Intelligence & Virtual Reality (IEEE AIVR)
In wearable AR/VR systems, data transmission between cameras and central processors can account for a significant portion of total system power, particularly in high framerate applications. Thus, it becomes necessary to compress video streams to reduce the cost of data transmission. In this paper we present a CNN-based compression scheme for such vision systems. We demonstrate that, unlike conventional compression techniques, our method can be tuned for specific machine vision applications. This enables increased compression for a given application performance target. We present results for Detectron2 Keypoint Detection and compare the performance and computational complexity of our method to existing compression schemes, such as H.264. We created a new high-framerate dataset which represents common scenarios for wearable AR/VR devices.
Daniel Haimovich, Dima Karamshuk, Thomas Leeper, Evgeniy Riabenko, Milan Vojnovic
Liqi Yan, Qifan Wang, Yiming Cu, Fuli Feng, Xiaojun Quan, Xiangyu Zhang, Dongfang Liu
Patrick Lewis, Barlas Oğuz, Wenhan Xiong, Fabio Petroni, Wen-tau Yih, Sebastian Riedel