The computational power of network nodes whether it is a mobile device or a miniaturized computer increases year by year, however the capacity of wireless channels in 4G networks is still falling behind for real-time video transmission. The paper proposes utilizing the computational capacities of mobile devices to detect objects of interest and apply less compression to corresponding regions, while the rest of the video frame is compressed with a higher compression ratio. We generalized this approach on multiple compression ratios depending on the importance of detected objects. For testing purposes we implemented a transmission system that takes into account the content that is being transmitted.
Paper
Full text
Video Transmission Using Multi-Level Content Aware Compression Based on Object Detection
Semantic Scholar · Computer Science · 2014
Abstract
The computational power of network nodes whether it is a mobile device or a miniaturized computer increases year by year, however the capacity of wireless channels in 4G networks is still falling behind for real-time video transmission. The paper proposes utilizing the computational capacities of mobile devices to detect objects of interest and apply less compression to corresponding regions, while the rest of the video frame is compressed with a higher compression ratio. We generalized this approach on multiple compression ratios depending on the importance of detected objects. For testing purposes we implemented a transmission system that takes into account the content that is being transmitted.