Mathieu Salzmann

École Polytechnique Fédérale de Lausanne, International Computer Science Institute

Papers

Total Citations

H-Index

About

Mathieu Salzmann is a leading researcher in computer vision and machine learning, with a primary focus on human motion prediction and multi-modal perception. His work addresses critical challenges in safety-critical applications such as human-robot interaction and autonomous driving. Salzmann’s major contributions include the development of the Temporal Inception Module for motion prediction, a novel architecture that effectively exploits different temporal scales for varying input lengths, significantly improving forecasting accuracy. His research also introduced the concept of using keyposes for long-term motion prediction, demonstrating that predicting every time instant is unnecessary and that focusing on key frames yields more robust and efficient results. With over 80 citations across his most-cited works, including the highly influential 2021 paper on the Temporal Inception Module (41 citations), Salzmann has made a lasting impact on the field. His earlier work on learning to recognize objects from unseen modalities (25 citations) further showcases his versatility in addressing fundamental problems in computer vision. Through these innovations, Salzmann continues to shape the future of autonomous systems and human-aware AI.