We are at the very beginning of the era of Internet of Video Things (IoVT), where many cameras collect a huge amount of visual data to be analyzed.  As the number of cameras and applications grows exponentially, it is critical to use artificial intelligence (AI) to process this data because humans cannot handle it all. However, designing efficient IoVT systems poses many challenges, such as accuracy, energy efficiency, and processing speed.

To address these challenges, this talk will focus on six topics related to the co-optimization of video compression and computer vision algorithms for efficient IoVT systems:
1. AI-aware compression, which optimizes video compression for machines to consume the video for better video analytics results, rather than optimizing for human perceptual comfort.
2. AI-assisted compression, which uses AI algorithms to assist the decisions made in compression tools for different types of video, such as video-on-demand vs. live broadcast.
3. AI-based compression, where the image/video is compressed using AI algorithms, such as deep learning, instead of commonly used standards.
4. Compression-aware AI, which ensures that computer vision algorithms are aware of the potential artifacts caused by lossy compression, improving the accuracy of the video analytics system.
5. Compression-assisted AI, which uses information from the compressed bit streams, such as motion vectors, to assist the computer vision algorithms.
6. Compression-based AI, which applies computer vision directly to compressed domain data, reducing the decompression time.

Overall, the co-optimization of video compression and computer vision algorithms is critical to designing efficient IoVT systems. This talk aims to shed light on the challenges and opportunities involved in this co-optimization process.

Yen-Kuang Chen received his Ph.D. degree from Princeton University. His research areas span from emerging applications that can utilize the true potential of multimedia and Internet of Things (IoT) to computer architecture that can embrace emerging applications. He has 100+ patents and 100+ technical publications. He is one of the key contributors to Supplemental Streaming SIMD Extension 3 and Advanced Vector Extension in Intel microprocessors. He has given 5+ keynote speeches, e.g., AICAS 2021, and 10+ tutorials at IEEE International Conferences, e.g., VCIP 2012 & 2011, ISCAS 2012 & 2009, and ICME 2010 & 2007. He is recognized as an IEEE Fellow for his contributions to algorithm-architecture co-design for multimedia signal processing.