Boisseau romain

International project in collaboration with DFKI in implementing a more robust monocular Visual SLAM in challenging environments. Mentorship to PhD students team. Research, design and integrate multiple high-level features through machine learning models: Objects recognition to ellipses/ellipsoids discretization, normal and plane map. Novel relocalization methods: restart from any position thanks to objects. Awarded demo at CVPR 2023. Plane reconstruction and tracking in real-time and AR demonstration.

Digital consultant on Unity (C#) focused on AR (Augmented reality) mobile applications. Expertise in integrating AR with AI technologies, including voice and object recognition, to create interactive experiences. Developed multiplayer AR environments, enabling users from different locations to interact seamlessly within the same virtual space. Handled occlusion challenges, ensuring realistic interaction between virtual objects and the real world. Experienced in using gestures for mesh customizations.

Developed and implemented deep learning algorithms for automated license plate recognition, including classification, localization, and reading, from scratch. Deployed the system into production, automating image processing for 68% of cases with a high accuracy rate of 1 error per 10,000 images.