Analysis of Traffic Condition by using Machine Vision
Nowadays, there is a large number of students and staffs at Suranaree University of Technology. Since vehicles are required for daily transportation therefore, the number of students and staffs is having a remarkable impact on the number of vehicles, traffic congestion, and parking availability within the university area. This research focuses on the analysis of traffic condition using machine vision in classifying and counting vehicle. Traffic videos were recorded by using an IP camera installed at the junction area (entrance) of the general inspection building (Building 1). Three types of vehicles were considered in this research: small, mid-size, and large vehicles. The collected videos were used in machine vision process to isolate the interested vehicle from the background and to detect its dimensions. The experimental results of vehicle counting and classification analyzed from 50 traffic videos show that the average accuracy (%) of small vehicles, mid-size vehicles and large vehicles are 87.51, 90.01 and 93.72 respectively.
Keywords - Detection; BackgroundSubtracktorMOG2; OpenCV