Indonesian Journal of Electrical Engineering and Computer Science Vol. 21, No. 3, March 2021, pp. 1474~1484 ISSN: 2502-4752, DOI: 10.11591/ijeecs.v21.i3.pp1474-1484  1474 Journal homepage: http://ijeecs.iaescore.com Accident vehicle types classification: a comparative study between different deep learning models Mardin Abdullah Anwer, Shareef M. Shareef, Abbas M. Ali Software and Informatics Department, College of Engineering, Salahaddin University-Erbil, Iraq Article Info ABSTRACT Article history: Received Sep 24, 2020 Revised Nov 27, 2020 Accepted Dec 16, 2020 Classifying and finding type of individual vehicles within an accident image are considered difficult problems. This research concentrates on accurately classifying and recognizing vehicle accidents in question. The aim to provide a comparative analysis of vehicle accidents. A number of network topologies are tested to arrive at convincing results and a variety of matrices are used in the evaluation process to identify the best networks. The best two networks are used with faster recurrent convolution neural network (Faster RCNN) and you only look once (YOLO) to determine which network will identifiably detect the location and type of the vehicle. In addition, two datasets are used in this research. In consequence, experiment results show that MobileNetV2 and ResNet50 have accomplished higher accuracy compared to the rest of the models, with 89.11% and 88.45% for the GAI dataset as well as 88.72% and 89.69% for KAI dataset, respectively. The findings reveal that the ResNet50 base network for YOLO achieved higher accuracy than MobileNetV2 for YOLO, ResNet50 for Faster RCNN with 83%, 81%, and 79% for GAI dataset and 79%, 78% and 74% for KAI dataset. Keywords: Accident recognition Deep learning GAI and KAI datasets Transfer learning Vehicle accidents image classification This is an open access article under the CC BY-SA license. Corresponding Author: Mardin Abdullah Anwer Software and Informatics department Salahaddin University- Erbil, Iraq Email: mardin.anwer@su.edu.krd 1. INTRODUCTION Intelligent transportation system has become one of the civilizational developments that the world is witnessing. Monitoring and analyzing the road has become quite necessary to control some areas instead of human beings-this breakthrough has brought into vision what is known as the smart city. Auto accident recognition and reporting system is a crucial topic; however, it has some problems such as multiple object tracking, object detection, and video surveillance in addition to some other issues related to the real-time monitoring system. The main objective of this research is to find the best deep learning network for describing vehicle accident images. Since use of roads by vehicles is on the increase, numerous studies on accident detection and classification are conducted utilizing signals such as acoustic where cross-correlation processing of radiated tire noise is used [1]. Radar [2] and ultrasonic signals [3] have been used to solve the problem of computational complexity in vehicle detection. Infrared thermal images taken from infrared thermal cameras are used to analyse road traffic flow surveillance in various circumstances, including conditions of poor visibility [4]. Many approaches using convolutional neural networks (CNNs) based on deep learning have been applied for vehicle classification and detection [5, 6]. At the time of the accident, vehicle data such as tag, picture, time, area, volume, and type are reported in written form by a police officer while in developed countries such information is passed on using intelligent transportation systems [7]. Research in this area has