Vehicle Type Classification and Counting Using YOLOv4 Algorithm

Abstract

The study implements a system that detects, classify and count vehicles based on their body type. Classifying and counting has proven to be beneficial when monitoring and managing traffics. However, there are few of studies that focuses on classifying and counting vehicles based on their car types. Implementing the YOLOv4 for classification and counting for the car types coupe, pickup, sedan, sports utility vehicle (SUV) and van, obtained an accuracy of 92.13% for classification and 89.14% for counting. The system was able to successfully classify and count vehicles based on their car type under one system compared to other system that only counts vehicles without classifying the car types.