การรู้จำยานยนต์จากภาพดิจิตอลโดยใช้โครงข่ายประสาทเทียม

Abstract

The topic of “Vehicle Recognition from Digital Image by Neural Networks” has the main purpose to study the application Artificial Neural Network to recognize the vehicle. Develop the program by Java. We train the Artificial Neural Network with images from 10 models 5 brands including Toyota Honda Mitsubishi Nissan Mazda In the process step. Find the center of wheels and extract the image to 2 parts, Front-wheel and Rear-wheel. Then, Analyze images in Artificial Neural Network. The results were satisfactory and the efficiency of the recognition, when train The Artificial Neural Network, The efficiency of the recognition and result depend on inputs and Training Images. The average accuracy by matching model Vehicle is about 87%