Research Article
A Comparative Study of YOLOv5, YOLOv8, Faster R-CNN, and Single Shot Detector (SSD) for Vehicle Detection in Indian Traffic Scenarios
@INPROCEEDINGS{10.4108/eai.19-12-2025.2366263, author={Sharwil Bhende and Shlok Anand and Akhil Gupta and Monali Gulhane and Firdous Sadaf Mohammad Ismail and Garima Shukla and Nitin Rakesh}, title={A Comparative Study of YOLOv5, YOLOv8, Faster R-CNN, and Single Shot Detector (SSD) for Vehicle Detection in Indian Traffic Scenarios}, proceedings={Proceedings of the 1st International Conference on Computational Techniques in Data Science, IICTDS 2025, 19-20 December 2025, Chandigarh, India}, publisher={EAI}, proceedings_a={IICTDS}, year={2026}, month={11}, keywords={Vehicle Detection Indian Traffic YOLOv8 Real-Time Processing Deep Learning Object Detection Edge Deployment mAP FPS Occlusion Handling}, doi={10.4108/eai.19-12-2025.2366263} }- Sharwil Bhende
Shlok Anand
Akhil Gupta
Monali Gulhane
Firdous Sadaf Mohammad Ismail
Garima Shukla
Nitin Rakesh
Year: 2026
A Comparative Study of YOLOv5, YOLOv8, Faster R-CNN, and Single Shot Detector (SSD) for Vehicle Detection in Indian Traffic Scenarios
IICTDS
EAI
DOI: 10.4108/eai.19-12-2025.2366263
Abstract
Indian streets are a lively mess consisting of cars, bikes, rickshaws, and trucks, all are struggling to find a place on sun-baked or rain-soaked roads. In this place of traffic, it is not only technically difficult to detect all vehicles at the same time and in the real world but it is the basis of smarter cities with safer roads. In this paper, we will compare four popular deep learning models: YOLOv5, YOLOv8, Faster RCNN, and SSD, which are applied to real-life video shots that can be found on Indian highways and urban roads. Mean Average Precision (mAP@0.5), inference speed (FPS), and ability to handle occlusions, lighting changes and small or overlapping objects are measured. Findings indicate that YOLOv8 has reached the right balance in a fast enough manner to deploy edges in real-time, and the right balance that is accurate in Indian dense traffic. This paper fills a fundamental gap in region-specific benchmarking and provides a useful roadmap to intelligent traffic systems specific to the road conditions of India.