Numerous deep learning programs have been developed related to the classification of vehicle groups. Still, none of them specifically applied to group-III, IV, and V, which are the groups that consist of trucks with similar models. This program combines both Web Scraping method to collect vehicles’ images as datasets, and YOLOv4 (You Only Look Once version 4) algorithm for classification of groups-III, IV, and V vehicles using python programming language, OpenCV library, and darknet framework. The dataset is collected from three different sources, which are Google Image, “Dataset of vehicle images for Indonesia toll road tariff classification”, and “Truck Image Dataset”. This dataset is split into 80% of training set, 15% of validation set, and 5% of testing set. Testing will use images from the 5% test set. Testing is done by evaluating the accuracy of test set’s images with three different weights files and 0.25 threshold. The three chosen weight files are the weight file at 1400, 3000, and 3290 iterations. The highest average accuracy is 81% from the third scenario which use the 3000 iterations weight file. Each class scores an average accuracy of 76% to 98.8%.