Max Robotics — Global Service Robot Directory + US Certification

About

Juha Ylioinas is a computer vision researcher whose work centers on image-based localization and camera relocalization—critical problems for enabling robots and augmented reality systems to understand their position in 3D space. His major contributions focus on advancing learning-based approaches for estimating camera pose from a single image. Ylioinas is best known for pioneering full-frame scene coordinate regression, a technique that uses convolutional neural networks to predict 3D world coordinates for every pixel in a query image, bypassing traditional feature matching. His 2018 paper on this method has accumulated 38 citations, establishing a foundation for subsequent work. He further refined this approach by introducing an angle-based reprojection loss, detailed in his 2019 publication (33 citations), which improved pose accuracy by optimizing geometric consistency rather than simple coordinate distance. This innovation addresses a key limitation in earlier regression models. Ylioinas’s research has direct applications in robotics, autonomous navigation, and augmented reality, where robust, real-time localization is essential. His work represents a significant step toward end-to-end learned localization systems that are both efficient and accurate.