Comparison of KNN, ANN, CNN and YOLO algorithms for detecting the accurate traffic flow and build an Intelligent Transportation System

Abstract

This main objective of this System is to perform object detection which serves as the foundation for controlling critical traffic data such as vehicle count, vehicle movements. This study compares state-of-the-art object detectors used to incorporate traffic state estimates. To distinguish vehicles, three different object detectors are compared. Here four algorithms are compared to show which algorithm is the best to detect the vehicle. The algorithms used are CNN, KNN, HAARCASCADE and YOLO. Algorithms which are considered as group 1 and 2 respectively. The sample size considered for implementing this work was <tex>$N=10$</tex> for each of the groups considered. The sample size calculation was done using clincalc. The pretest analysis was kept at 80&#x0025;. During the process of testing five different videos were taken and the accuracy rate was calculated and the accuracy rate appears to be high for YOLO with 93&#x0025; which is compared with three algorithms such as KNN, Haarcascade and CNN algorithm with accuracy of 59&#x0025;, 89&#x0025; and 59&#x0025; respectively and proved that YOLO performance better in novel vehicle detection in real time road traffic. The statistical significance value achieved for calculation of accuracy was found to be 0.187The effectiveness of the YOLO object detection looks to be superior than CNN, KNN, and HAARCASCADE based on the accuracy value.