Yangxintong Lyu received the MSc. degree in Applied Computer Science from Vrije Universiteit Brussel, Belgium, in 2018 and the B.Eng. degree in Software Engineering from Xidian University, China, in 2017. She is currently pursuing her PhD at ETRO supervised Adrian Munteanu.
Traffic scene understanding of image content has been used in computer vision broadly, which is an important ingredient for autonomous driving and traffic surveillance. Assigning semantic labels to pixels or detection object 2D bounding boxes have been well explored and achieved very promising results. However, in autonomous driving and traffic surveillance scenarios, more high-level and fine-grained representations of image content are needed for tasks like 3D spatial reasoning.
The main objective of my research is to propose Deep-Learning (DL) based algorithms for 3D representations from 2D/3D information under traffic scene. The output 3D representations can be further more used in real-world applications like vehicle 3D location prediction, or traffic scenarios simulation.