Exploiting Deep Visual Geometry Group Architecture for Fall Detection in the Elderly People

Abstract

Over the last couple of decades, human fall detection has gained considerable popularity, especially for the elderly. Elderly people need more attention as compared to others in their homes, hospitals, and care centers. Various solutions have been proposed to deal with this problem, yet, many aspects of this problem are still unresolved. The current study proposed an approach for human fall detection based on the Visual Geometry Architecture of deep learning. The presented approach was weighed up with state-of-the-art approaches including ResNet-50 and even ResNet-101 by using MCF and URFD datasets, outperforming them with an accuracy of 98%. The proposed approach also outperformed these deep architectures in terms of performance efficiency.