GOLLA. KHYATHI, PROF. G SREENIVASULU

DOI: https://doi.org/10.5281/zenodo.23346883

With the number of vehicles growing at an accelerated rate, efficient traffic monitoring and vehicle detection systems are needed. The current methods of traffic monitoring are largely manual, thus requiring lots of time and manpower. In this paper, a YOLOv4 object detection algorithm is used to develop an optimized real-time vehicle detection framework for video streams. The proposed system receives image and video inputs, and outputs vehicle detection results in the form of a bounding box, class label and confidence score. The system offers individual modules for detection of images and live stream vehicles. The object detection is performed by YOLOv4 for fast and efficient detection. The developed framework can be adopted for traffic surveillance, vehicle monitoring and intelligent transportation applications. The experimental results show the vehicle detection capabilities of the system from traffic images and live video streams.

Vehicle Detection, YOLOv4, Deep Learning, Object Detection, Computer Vision, Video Surveillance, Real-Time Detection

A real-time vehicle detection framework optimized with the deep learning based YOLOv4 object detection algorithm was presented in this paper. The proposed system will detect vehicles from both static images and prerecorded videos as well as live video stream. Using a pretrained YOLOv4 model helps the system classify the commonly used vehicle types without starting to train the model again. The proposed framework processes the input frames and identifies vehicles by drawing bounding boxes, labeling vehicle types and calculating vehicle confidence scores. The frames are processed continuously for video input, to generate results of vehicle detection in real-time. The system also offers independent image-based and live-stream detection, enabling the framework to be applied to various traffic monitoring needs. The developed system is applied to vehicle detection in traffic scenes as an example of the practical application of YOLOv4. The output frames contain information about the location and class of detected vehicles which can be clearly visualized. The framework can then be applied to different applications like traffic surveillance, vehicle monitoring, road safety, and intelligent transportation systems. Proposed system is efficient in real-time vehicle detection, but the detection performance is not guaranteed in the case of low light, high traffic density, vehicle occlusions, motion blur, complex background and bad weather. They can impact vehicle visibility and appearance and can decrease detection reliability. In the future, there are opportunities to improve the accuracy of detection and processing speed for challenging environmental conditions. Vehicle tracking, vehicle counting, speed estimation, real-time traffic analysis can also be added to the system to be extended. The system can be deployed to edge devices and other resource-constrained devices to enhance system usage and utility in the field. In general, the proposed YOLOv4 based framework serves as a good basis for fast vehicle detection system and can be further improved for other advanced intelligent transportation and traffic surveillance applications

[1] R. Wang et al., “Real-time vehicle target detection in inclement weather conditions based on YOLOv4,” Frontiers in Neurorobotics, vol. 17, 2023.
[2] P. Kumeechai, “Real-time vehicle detection system on the highway,” Journal of Applied Research on Science and Technology, 2023.
[3] S. Mishra and D. K. Yadav, “Vehicle Detection in High Density Traffic Surveillance Data using YOLO.v5,” 2023/2024.
[4] “A framework for real-time vehicle counting and velocity estimation using deep learning,” Sustainable Computing: Informatics and Systems, vol. 40, 2023.
[5] X. Liu et al., “YOLOv8-FDD: A Real-Time Vehicle Detection Method Based on Improved YOLOv8,” IEEE Access, vol. 12, 2024.
[6] D. Yu et al., “Real-Time Monitoring Method for Traffic Surveillance Scenarios Based on Enhanced YOLOv7,” Applied Sciences, vol. 14, 2024.
[7] Y. Li, Y. Huang, and Q. Tao, “Improving real-time object detection in Internet-of-Things smart city traffic with YOLOv8-DSAF,” Scientific Reports, vol. 14, 2024.
[8] Z. Cui et al., “RTL-Net: Real-time lightweight Urban traffic object detection algorithm,” Complex & Intelligent Systems, vol. 11, 2025.
[9] A. Hunain, I. Indrabayu, and A. A. Ilham, “Vehicle Detection and Tracking Techniques Using Yolo Detection and Kalman Filter Optimization,” 2025.

GOLLA. KHYATHI, PROF. G SREENIVASULU (2026). AN OPTIMIZED REAL-TIME VEHICLE DETECTION FRAMEWORK FOR VIDEO STREAMS USING YOLOv4. International Journal of Engineering and Techniques, 12(5), 350–361. ISSN: 2395-1303. DOI: https://doi.org/10.5281/zenodo.23346883