Alexandre Boulch

Papers

Total Citations

H-Index

About

Alexandre Boulch is a leading researcher in computer vision and robotics, specializing in 3D scene understanding, semantic segmentation, and depth estimation. His work bridges the gap between model-based and data-driven approaches, enabling robots to perceive and interact with complex environments more reliably. Boulch is best known for developing SnapNet-R, a pioneering framework for consistent 3D multi-view semantic labeling that integrates geometric coherence into deep learning pipelines—a critical advancement for autonomous navigation and mapping. This work has garnered over 80 citations, reflecting its influence on the robotics community. In parallel, Boulch has contributed to stereo vision with a novel method for refining disparity maps by fusing traditional model-based estimations with modern deep learning outputs, achieving robust real-time performance for applications in autonomous driving. His research consistently emphasizes practical, deployable solutions that balance accuracy with computational efficiency. Boulch’s achievements include advancing the state of the art in 3D perception, with his work cited by researchers developing next-generation robotic systems and autonomous vehicles. For students and researchers, his career demonstrates the power of combining classical geometric reasoning with contemporary machine learning to solve real-world perception challenges.