Xingqi Zou

Institute of Computing Technology, Chinese Academy of Sciences

Papers

Total Citations

H-Index

About

Xingqi Zou is a leading researcher in energy-efficient computer vision hardware, with a primary focus on stereo vision acceleration for real-time embedded systems. His most impactful contribution is the development of the Dadu-Eye stereo vision accelerator, which achieves an impressive 5.3 TOPS/W efficiency while processing 30 fps at 1080p resolution—a breakthrough that enables high-accuracy depth estimation on power-constrained platforms like drones and robots. This work, which has garnered 19 citations, uniquely combines lightweight deep neural networks with cost volume algorithms to balance speed and precision. Zou further advanced the field with Dadu-SV, demonstrating how stereo vision processing can be efficiently deployed on neural processing units (NPUs). His research addresses the critical challenge of making binocular vision systems—essential for autonomous vehicles, robotics, and AR devices—practical for real-world deployment by overcoming the computational bottlenecks of both classic semi-global matching and deep CNN approaches. Through these contributions, Zou has established himself as a key innovator in bridging the gap between algorithmic advances and hardware-efficient implementation, making high-performance stereo vision accessible for next-generation intelligent systems.