Apriandy Angdresey, Lanny Sitanayah, Efraim Pantas
The number of vehicles on the road increases from year to year. This causes road infrastructure to be frequently upgraded to meet the needs of road users. To find out the number of a kind of vehicle, we have to classify moving objects on the road. Convolutional Neural Network is one of the deep learning techniques for image processing and pattern recognition. It can be used to classify moving objects on the roads. In this paper, we implement and compare three Convolutional Neural Network architectures to classify moving objects from traffic CCTV videos. The three architectures are SSD MobileNet, RetinaNet, and EfficientNet. The objects that we try to classify are cars, motorcycles, trucks, and people. Our simulation results show that these three architectures can perform good detection and classification with an average accuracy above 83.4%. However, there is still room for improvement in small-area detection in the videos. © 2023 IEEE.
Universitas Katolik de la Salle, Dept. of Informatics Engineering, Manado, Indonesia
