Bicycles are low-carbon emitting and inexpensive means of transportation that are popular worldwide. In densely populated areas, bicycle sharing programs are helping to expand access to these vehicles. As bicycle sharing programs grow and more people use bicycles on the road, there is a need to improve the performance and safety of bicycle operation. One opportunity for improving the safety of bicycle operation is to add automated road sign detection and identification. Once a road sign is detected and identified, this data can be used to improve autonomous navigation of the vehicle, to share data between bicycles on the road for better traffic control, and to remind or prompt the operator about current road status. The Mask-RCNN deep learning architecture is trained to detect, identify, and segment several common Vietnamese road signs. Training images were captured from a camera mounted on a bicycle traveling on the road. Results show that the trained neural net successfully recognizes signs, with varying success rates across trials correlating heavily with the number and quality of training annotations.
Road Sign Segmentation for Smart Bicycles
Lect. Notes in Networks, Syst.
The International Conference on Intelligent Systems & Networks ; 2021 ; Hanoi, Vietnam March 19, 2021 - March 19, 2021
2021-05-13
8 pages
Aufsatz/Kapitel (Buch)
Elektronische Ressource
Englisch
Deep learning , Mask RCNN , Internet of things , Bicycle , Transportation , Artificial intelligence Engineering , Computational Intelligence , Security Science and Technology , Data Mining and Knowledge Discovery , Cyber-physical systems, IoT , Professional Computing , Signal, Image and Speech Processing
Engineering Index Backfile | 1944
|