Yingwei Li

Papers

Total Citations

H-Index

About

Yingwei Li is a computer vision researcher whose work focuses on the security and robustness of deep neural networks, particularly in the context of 3D scene understanding. His most cited paper, "Adversarial Attacks on Monocular Depth Estimation" (2020), with 14 citations, investigates the vulnerability of depth estimation models—a critical component for autonomous driving and robotics—to carefully crafted adversarial perturbations. This research highlights how even state-of-the-art deep learning systems can be fooled, raising urgent concerns for real-world deployment in safety-critical applications. Li’s contributions lie at the intersection of adversarial machine learning and geometric computer vision, demonstrating that attacks on 2D images can propagate to disrupt 3D perception tasks. His work has helped catalyze a growing research area focused on defending depth estimation models against malicious inputs. By exposing these weaknesses, Li provides essential insights for building more trustworthy AI systems, making his research highly relevant for students and engineers working on robust vision pipelines.