Our research interests are in the areas of visual information processing that considers the entire video pipeline -- capture, compression, communication, and consumption. Our work also addresses enabling and oprimizing visual information processing targeting new modalities of presentation such as AR/VR. One key area of focus is visual information processing targeting machine consumption. With a machine (computer vision system) as the end consumer, we reimagine video representation and compression. A machine does not care about pixel fidility, just feature fidility necessary to perform a visual computing task.
We are actively involved in research related to audio-visual compression and communication standards. Our past work includes low complexity encoding and transcoding of HEVC, AVC/H.264, and MPEG-2. I am a co-inventor of 14 patents related to HEVC, AVC/H.264, ATSC, and Blu-ray standards and were licensed to manufacturers through patent pools such as the MPEG LA. Current efforts are focused on Versatile Video Coding (VVC/H.266) standard. Post standardization activities related to this work include VVC complexity reduction and energy-aware coding. We are actively participating in ISO/MPEG standardization efforts related to Video Coding for Machines (VCM).
Selected Research Projects