student profile: Ms Wei Zhou


Map

Thesis work

Thesis title: Semantic Segmentation for Intelligent Vehicles

Supervisors: Eduardo NEBOT , Stewart WORRALL

Thesis abstract:

Over the past decade, vision-based urban scene recognition has shown its capability of providing rich information to intelligent transportation systems. Semantic segmentation, as a pixel-level scene parsing method, can provide high-level understandings from images of urban scenes. However, the commonly used Convolution Neural Network (CNN) features are biased to the specific datasets and do not adapt well to a specific environment. Therefore, this research is going to explore the transferability of CNN features and adaptation of pre-trained models to local scenarios. By fusing the segmented information with several sensors, the electrical vehicles can make better control decision for autonomous driving.

Note: This profile is for a student at the University of Sydney. Views presented here are not necessarily those of the University.