Traffic accidents are caused by driver faults and non-compliance with traffic rules mostly. It is aimed to reduce the accident rates by preventing driver-related accidents was the developments in electric and autonomous vehicle technologies by the studies carried out in recent years. Lane tracking systems for autonomous vehicles are critical to reducing these kinds of risks in accidents. Within the scope of this study, a test environment was created in the Gazebo environment for lane detection and tracking. In addition, segmentation process and object detection algorithms were used for lane detection by utilizing Mask RCNN and Faster RCNN. The software developed within the scope of the study has been successfully tested in the created simulation environment.
Lane Tracking with Deep Learning: Mask RCNN and Faster RCNN
07.09.2022
399218 byte
Aufsatz (Konferenz)
Elektronische Ressource
Englisch
Preprocessed Faster RCNN for Vehicle Detection
IEEE | 2018
|Improved Faster RCNN for Traffic Sign Detection*
IEEE | 2020
|A Faster RCNN-Based Pedestrian Detection System
IEEE | 2016
|Robustness of Mask-RCNN at High Ambiguous Environments
IEEE | 2022
|