Human Fall Analysis with FallAllD Dataset using Machine Learning Approach

The growing occurrence of falls among elderly individuals has garnered substantial attention from healthcare experts and researchers. Ensuring the prevention of fall-related injuries and fatalities necessitates precise and streamlined fall-detection mechanisms. This study introduces an innovative strategy for multi-class fall detection through the utilization of machine learning techniques in conjunction with the FallAllD dataset. The outcomes of the evaluation underscore the efficacy of the proposed method and its potential to enhance fall detection within real world scenarios. The CNN based 5 class classifier is able to detect fall with 99.24% accuracy. Investigation of data collected from sensor boards connected with waist and hand were analyzed.