Content-Adaptive Video Retargeting
Content-Adaptive Video Retargeting

This project aims to develop a complete system for delivering high-quality stereoscopic broadcast video. We focus on the real-time video of sporting events and soccer in particular. The system has a number of challenging components. First, the video needs to be automatically converted from 2D to stereoscopic 3D content. This task needs to be carried out in near real-time. Second, the stereoscopic video needs to be adapted for each display type and streamed to the client (e.g., a mobile phone, tablet, or VR headset). Third, a client software needs to be designed and developed to track a viewer and to interactively adjust the 3D content. In addition, we develop supporting algorithms and systems that take operate at the intersection of computer vision, image processing, and machine learning.

Principal Investigators
Mohamed Elgharib, QCRI
Wojciech Matusik, CSAIL