SEATBELT DETECTION IN TRAFFIC SYSTEM USING AN IMPROVED YOLOv5
Abstract
Seat belt use is essential for avoiding fatalities and serious injuries in car accidents. Due to the challenge of identifying vehicles in images of a traffic scene that are affected by complicated illumination, the typical seat belt identification algorithm has low accuracy for determining the driver's seatbelt status. We proposed seat belt detection in a traffic management system utilizing an improved YOLOv5 algorithm developed by combining YOLOv5 with brightness augmentation. This procedure is used to improve existing algorithms for seat belt detection. The framework incorporates image enhancement, region proposal generation, depth feature extraction, target recognition, and detection into a convolutional neural network model, which significantly boosts training efficiency and detection accuracy. To examine the performance of the upgraded YOLOv5 detection algorithm, a benchmark dataset known as the Yawning Detection Dataset was gathered. The examination focuses on identifying the state of a safety belt between two classes: "seatbelt" and "non-seatbelt". The results demonstrate a high level of accuracy, with a mean average precision (mAP) of 96%, Precision of 99%, Recall of 99%, and a true positive score of 95.7%, indicating the system’s effectiveness in identifying the safety belt status.
Authors who publish with this journal agree to the following terms:
- Authors retain copyright and grant the journal right of first publication with the work simultaneously licensed under a copyright form (JACTA) that allows others to share the work with an acknowledgement of the work's authorship and initial publication in this journal.
- Authors are able to enter into separate, additional contractual arrangements for the non-exclusive distribution of the journal's published version of the work (e.g., post it to an institutional repository or publish it in a book), with an acknowledgement of its initial publication in this journal.
- Authors are permitted and encouraged to post their work online (e.g., in institutional repositories or on their website) prior to and during the submission process, as it can lead to productive exchanges, as well as earlier and greater citation of published work (See The Effect of Open Access).