Jiajun Li

Institute of Computing Technology, University of California, Santa Barbara, Chinese Academy of Sciences

Papers

Total Citations

H-Index

About

Jiajun Li is a researcher at the forefront of energy-efficient computer vision and deep learning acceleration for embedded and mobile systems. His work primarily focuses on enabling high-performance 3D perception—including stereo vision and point cloud processing—on resource-constrained platforms like robots and drones. Li’s major contribution is the development of specialized hardware accelerators that bridge the gap between algorithmic complexity and real-time, low-power deployment. His most cited work, "Dadu-Eye," presents a 5.3 TOPS/W stereo vision accelerator achieving 30 fps at 1080p resolution, demonstrating how lightweight neural networks combined with cost-volume algorithms can deliver both high accuracy and speed. This innovation has garnered 19 citations, reflecting its impact on practical robotics and autonomous systems. Li has also applied deep convolutional neural networks to marine biology, developing a fish image classification system (16 citations) that aids in species monitoring and conservation. Additionally, his research on accelerating DNN-based 3D point cloud processing (11 citations) addresses the critical challenge of mobile computing efficiency. Through these contributions, Li is shaping the future of intelligent, low-power vision systems for real-world applications.