REAL-TIME DETECTION OF TRAFFIC SIGNS WITH YOLO ALGORITHMS
Author:
Open Access
- 1 March 2025
- journal article
- Published by Konya Muhendislik Bilimleri Dergisi in Konya Journal of Engineering Sciences
- Vol. 13 (1) , 220-237
Abstract
The number of electric vehicles is increasing day by day. The biggest reason for the increase in electric vehicles is their autonomous or semi-autonomous use feature. Autonomous or semi-autonomous driving; It is the movement of the vehicle with the data coming from the sensors, cameras, and sensors around the vehicle. The majority of traffic accidents are caused by driver errors. The most important of these mistakes is not obeying traffic rules. Autonomous or semi-autonomous driving largely prevents driver-related traffic accidents. The biggest problem of autonomous vehicles is the difficulties in detecting traffic signs in real-time. The locations, shapes, and scales of traffic signs are very different. Traffic signs are difficult to detect in real-world conditions due to their similarity to other objects. The study carried out real-time detection of traffic signs. For this purpose, images were taken from the camera placed inside the vehicle. A data set was created with these images. The more real environment images the data set consists of, the more accurate the real-time detection process increases. In this study, 8931 traffic sign images were taken from real environments. These images were taken from different locations, different lighting levels, and different distances. In addition, the number of data was increased to 78895 by adding grayscale, adding slope, blurring, adding variability, adding noise, changing image brightness, changing colour vividness, changing perspective, resizing, and positioning the images. With this study, the data set was adapted to the real environment. The created data set was used in 3 different versions of YOLOv5 architecture, YOLOv6, YOLOv7 and YOLOv8 architectures. As a result of the study, the highest accuracy was found to be 99.60%, F1-Score was 0.962 and mAP@.5 value was 0.993 in YOLOv8 architecture.Keywords
This publication has 47 references indexed in Scilit:
- Evolving traffic sign detection via multi-scale feature enhancement, reconstruction and fusionDigital Signal Processing, 2025
- TSD-DETR: A lightweight real-time detection transformer of traffic sign detection for long-range perception of autonomous drivingEngineering Applications of Artificial Intelligence, 2024
- Improved YOLOv8 for small traffic sign detection under complex environmental conditionsFranklin Open, 2024
- EDN-YOLO: Multi-scale traffic sign detection method in complex scenesDigital Signal Processing, 2024
- Mexican traffic sign detection and classification using deep learningExpert Systems with Applications, 2022
- Traffic sign recognition based on deep learningMultimedia Tools and Applications, 2022
- A real-time vehicle detection and a novel vehicle tracking systems for estimating and monitoring traffic flow on highwaysAdvanced Engineering Informatics, 2021
- Underwater target detection based on Faster R-CNN and adversarial occlusion networkEngineering Applications of Artificial Intelligence, 2021
- Will people accept shared autonomous electric vehicles? A survey before and after receipt of the costs and benefitsEconomic Analysis and Policy, 2019
- Robust Lane Detection Based On Convolutional Neural Network and Random Sample Consensus2014