Comparison of YOLO Algorithms for Vehicle Accident Detection and Classification
Nusari, Ahmed N., Alawi, Abdulfattah E. Ba, Bozkurt, Ferhat, Ozbek, Ibrahim Yucel
Early detection of vehicle accidents is essential for immediate response and to decrease the severity of injuries while vehicle accidents occur. Among the advanced algorithms YOLOv6, YOLOv7, YOLOv8, and YOLOv9 are compared in this article to determine their performance in detecting vehicle accidents that are categorized as "severity" or "moderate." The study emphasizes how important deep learning is to enhancing accident detection systems. Out of all the models evaluated, YOLOv8m has the highest precision (0.979), and the highest recall (0.927) for YOLOv8n. Furthermore, YOLOv9c performs admirably in mAP50, receiving a score of 0.977, indicating its resilience in identifying accidents under a variety of circumstances. YOLOv9e is noteworthy for its better mAP5095, which reaches a high of 0.941, demonstrating its ability to localize accident incidents accurately. These results provide valuable insight into how well particular algorithms perform in real-world situations. Through a close examination of their performance measures, this research aids in the advancement of more effective accident detection systems, which in turn improves emergency response operations and lessens the effects of accidents on people and communities. This comparison research emphasizes how crucial it is to use state-of-the-art technology to increase accident detection accuracy and enable timely and efficient response.
- Dergi
- 2024 4TH INTERNATIONAL CONFERENCE ON EMERGING SMART TECHNOLOGIES AND APPLICATIONS, ESMARTA 2024
- Yıl
- 2024
- Sayfa
- 493-500
- DOI
- 10.1109/eSmarTA62850.2024.10638929
- WoS UID
- WOS:001308291300068