Vision-Based Fall Detection Systems Using 3D Skeleton Features for Elderly Security: A Survey

In recent years, the pace of the elderly people population increasing dramatically every day and even more than before. For elderly people, falls are one of the most critical healthcare risks and have serious consequences that may lead to death. Therefore, the fall detection system has become an important research topic in the medical and health-care fields. Fall detection based on skeleton data is considered a good solution to timely feedback and overcome the risk caused by falls. Skeleton data especially of 3-dimensional space is not affected by a popular occlusion. In recent years, numerous studies have proven that combining 3D skeleton datasets with deep learning has its unique advantages. In this survey, we first introduce the fall detection techniques categories, then we introduce the process of fall detection machine learning model and the commonly used 3d skeleton data fall detection. Finally, we look closely at a comparative investigation of existing variants of approaches in vision-based fall detection tasks from different perspectives.