Intelligent Transportation System (ITS) aims at improving the safety on roads, and with the advancement in vehicular technology, autonomous vehicles will soon be equally present on roads along with manual vehicles. Vehicle classification is used by autonomous vehicles to decide driving behavior. In this paper, we use transfer learning in deep learning for vehicle classification. Different algorithms are implemented and compared using Kaggle dataset. It is evident from the experimental results that the hybrid InceptionResNet algorithm exhibits the highest training accuracy of 92 percent and validation accuracy of 62 percent. Similarly, the validation of the InceptionResNet model results in 85 percent AUC. The classification results can be used by autonomous vehicles to improve driving behavior like lane changing, overtaking, etc., and improve the safety of the commute.
Vehicle Classification for Autonomous Vehicles Using Transfer Deep Learning
Lect. Notes Electrical Eng.
International Conference on Signals, Machines, and Automation ; 2022 ; Delhi, India August 05, 2022 - August 06, 2022
2023-05-23
7 pages
Aufsatz/Kapitel (Buch)
Elektronische Ressource
Englisch
Deep Learning & Autonomous Vehicles
TIBKAT | 2018
|Motion Prediction for Autonomous Vehicle using Deep Learning Architecture and Transfer Learning
DOAJ | 2023
|