Max Robotics — Global Service Robot Directory + US Certification

About

Yihui He is a computer vision researcher whose work sits at the intersection of object detection, visual tracking, and efficient deep learning — with a particular focus on making AI systems safer and more deployable in real-world robotics and autonomous driving applications. His most recognized contribution, the Deep Mixture Density Network for Probabilistic Object Detection, addresses a critical challenge in robotics: quantifying uncertainty in object localization to reduce the risk of catastrophic failures in deployment scenarios such as occlusion-heavy environments. This work has accumulated 17 citations and reflects He's commitment to building trustworthy perception systems. Complementing this, his research on motion prediction in visual object tracking (8 citations) offers a computationally efficient alternative to feature-heavy tracking pipelines, making robust tracking more accessible for resource-constrained platforms. More recently, He has explored neural network efficiency through depth-wise decomposition of separable convolutions, targeting the inference latency bottlenecks that limit the deployment of deep CNNs on edge devices. Across his portfolio, He consistently bridges theoretical innovation with practical system constraints, making him a notable contributor to the growing field of efficient and reliable computer vision.