Deteksi Seatbelt Pada Pengendara Mobil Melalui CCTV Menggunakan Metode YOLOv7
Abstrak
The use of seatbelts is crucial for vehicle safety, yet many drivers still neglect it, increasing the risk of injuries and fatalities. This research develops a real-time detection system for seatbelt usage using CCTV footage and the YOLOv7 (You Only Look Once version 7) algorithm. The method involves capturing traffic video, processing it into image frames, and detecting seatbelt usage with high accuracy. Data collection was conducted in several strategic locations in Surabaya, followed by preprocessing and annotation for object detection. The best model was achieved with an 80:20 data split, resulting in a mean Average Precision (mAP) of 0.964, Precision of 0.949, and Recall of 0.949, demonstrating high accuracy and reliability. This system is expected to support traffic law enforcement and enhance safety monitoring through real-time detection
Identifier
- DOI: https://doi.org/10.33050/cerita.v12i13882
- OpenAlex: https://openalex.org/W7161638050