Improved Automatic Drowning Detection Approach with YOLOv8

Abstract

Although swimming is a popular activity that promotes relaxation and stress relief, drowning remains a serious global problem. According to the World Health Organization (WHO), drowning is the third most common cause of death. This study delves into implementing deep learning techniques for precise drowning detection. From this point of view, a drowning detection system was designed using the YOLOv8 model, which is a powerful tool for object detection and classification tasks. Using a large dataset, the YOLOv8 model was trained to recognize drowning patterns and movements and increase the likelihood of successful rescue operations by reducing response times and improving water safety. The proposed system uses deep learning techniques and YOLOv8 technology with data augmentation techniques to enhance the model's robustness to variations in lighting, pose, and background conditions. The system performance was evaluated using the Swimming and Drowning Detection dataset achieving 90.1% accuracy compared to 88.5% with YOLOv5.