This paper presents a fully automatic detection approach of vehicles using a deep neural network. We propose to explore the use of recent techniques of object detection and recognition. Firstly, we improve the quality of an image by correction and elimination of irrelevant information. Secondly, we properly prepare the training and validation set through a waterfall method. Thirdly, we apply a Fourier's transformation to extract object descriptors for representing the feature of the vehicle object. Finally, we adopt deep learning method to localize and recognize the vehicle model objects. To do this, deep neural network is training several iterations by machine learning methods. The object of this phase is to recognize only the vehicle from among all the other objects. Experimental result proves that our approach can effectively detect a vehicle with a good accuracy and recognition rate.
A Vehicle Detection Approach Using Deep Learning Network
2019-12-01
673576 byte
Conference paper
Electronic Resource
English
Anomaly Vehicle Detection Using Deep Neural Network
Springer Verlag | 2021
|Moving vehicle detection using deep neural network
IEEE | 2017
|Vehicle Number Plate Detection using Deep Learning
IEEE | 2024
|