Datasets for Autonomous Driving
Title: Datasets for Autonomous Driving
DNr: Berzelius-2023-50
Project Type: LiU Berzelius
Principal Investigator: Georg Hess <>
Affiliation: Chalmers tekniska högskola
Duration: 2023-03-31 – 2023-07-01
Classification: 10207


Deep learning has become a cornerstone in surround sensing systems in autonomous vehicles, and a core component of creating safe and robust autonomous driving is the use of high quality data. In this project, several novel types of neural networks are to be tried for different tasks in such surround sensing systems, and evaluated for different open datasets. The primary research questions revolve around processing videos in order to capture the dynamics of the scene or to improve performance for tasks typically tackled with single images. The project is a collaboration between members at different universities: Lund, Linköping, and Chalmers. Each member is part of another project with GPU-hours, but we apply for this project to have a common data storage and avoid storing the same dataset multiple times in different projects. This project is a continuation of Berzelius-2022-129, but with a new PI.