Vehicle analysis is an important and challenging process in the intelligent realization of applications, which includes the following sub-tasks, Vehicle Type Classification (VTC), License Plate Recognition (LPR), vehicle make and model recognition etc. When it comes to the 6 tasks, identification of vehicle models is integrally connected with LPR and helps improve the performance of intelligent systems. In this paper, a new solution based on the CNNs for detecting moving cars and model recognition is introduced. The framework to begin with identifies and extracts frontal-view images of vehicles, normalizes input data and key features for Model Recognition. These images are then employed in training as well as in testing CNNs, which are particularly suitable for learning complex features within an image. Our framework achieves outstanding recognition accuracy in the experiments performed, contributing to the framework’s possible application in the accurate identification of vehicle make and model and the development of vehicle analysis in smart systems. In this paper, the possibility of recognizing car models is the only examined application.
Car Model Recognition Using Convolution Neural Networks
06.09.2024
1048776 byte
Aufsatz (Konferenz)
Elektronische Ressource
Englisch