Max Robotics — Global Service Robot Directory + US Certification

About

Xiaotian Li is a leading researcher in computer vision and robotics, specializing in image-based localization and camera relocalization—the critical task of estimating a camera’s position and orientation from a single image. Li’s major contributions center on advancing deep neural network approaches for scene coordinate regression, directly predicting 3D world coordinates from image pixels to enable robust pose estimation. Their seminal 2018 work, “Full-Frame Scene Coordinate Regression for Image-Based Localization” (38 citations), pioneered end-to-end learning frameworks that replaced traditional feature-matching pipelines. Li further refined this with the innovative “Angle-Based Reprojection Loss” (2019, 33 citations), which improved geometric consistency during training by penalizing angular errors. Most recently, their 2024 paper “HSCNet++” (23 citations) introduces hierarchical scene coordinate classification and regression enhanced by Transformer architectures, pushing the state-of-the-art in visual localization. Li’s research has profound implications for autonomous vehicles, mobile robotics, and augmented reality, where reliable camera pose estimation is essential. With a consistent record of high-impact publications and a focus on bridging learning-based methods with geometric principles, Li continues to shape the future of visual localization technology.