A Vehicle Detection Approach using Deep Learning Methodologies

Abdullah Asim Yilmaz,Mehmet Serdar Guzel,Iman Askerbeyli,Erkan Bostanci

DOI: https://doi.org/10.48550/arXiv.1804.00429

2018-04-02

Abstract:The purpose of this study is to successfully train our vehicle detector using R-CNN, Faster R-CNN deep learning methods on a sample vehicle data sets and to optimize the success rate of the trained detector by providing efficient results for vehicle detection by testing the trained vehicle detector on the test data. The working method consists of six main stages. These are respectively; loading the data set, the design of the convolutional neural network, configuration of training options, training of the Faster R-CNN object detector and evaluation of trained detector. In addition, in the scope of the study, Faster R-CNN, R-CNN deep learning methods were mentioned and experimental analysis comparisons were made with the results obtained from vehicle detection.

Computer Vision and Pattern Recognition

What problem does this paper attempt to address?

The problem that this paper attempts to solve is to improve the success rate and efficiency of the detector in vehicle detection. Specifically, the author uses two deep - learning methods, R - CNN and Faster R - CNN, to train the sample vehicle data set, aiming to optimize the performance of the trained detector on the test data through these methods, so as to obtain more efficient vehicle detection results. The paper also compares the experimental analysis results of these two methods and evaluates their performance differences in vehicle detection tasks.

The main objectives of the paper can be summarized as follows:

1. **Successfully train vehicle detectors**: Use R - CNN and Faster R - CNN, two deep - learning methods, to train vehicle detectors on the sample vehicle data set.
2. **Optimize the detection success rate**: By testing the trained vehicle detectors and providing effective detection results, optimize the success rate of the detectors.
3. **Compare the performance of different methods**: Through the experimental analysis of R - CNN and Faster R - CNN methods, compare their performance differences in vehicle detection tasks, especially the detection quality, the mean average precision (mAP) value and the detection speed.

Through the achievement of these objectives, the paper aims to provide an efficient and accurate solution for the vehicle detection field.