Yilun Wu

Papers

Total Citations

H-Index

About

Yilun Wu is a researcher specializing in event-based vision and optical flow estimation, with a focus on developing efficient algorithms for resource-constrained robotic systems. His primary contributions lie in advancing lightweight neural network architectures that process event data—asynchronous streams of pixel-level brightness changes—to estimate motion without the heavy computational overhead of traditional frame-based methods. Wu’s most cited work, “Lightweight Event-based Optical Flow Estimation via Iterative Deblurring” (2024, 17 citations), introduces a novel approach that avoids expensive correlation volume construction, achieving state-of-the-art accuracy while drastically reducing compute and memory demands. This innovation is particularly impactful for drones, autonomous vehicles, and other edge devices where power and processing budgets are limited. An earlier version of this work (2022, 3 citations) laid the groundwork, demonstrating iterative deblurring as a key technique. By bridging the gap between high-performance event-based vision and practical deployment, Wu’s research enables faster, more efficient motion perception in real-world robotics, making him a notable contributor to the growing field of neuromorphic computing and efficient AI.