Fall Detection Technique for Older Individuals based on Deep Layered Neural Networks Embedded with Transfer Learning
J. Deepika, J. Akilandeswari
Abstract
The vital priority of the society is to render helping hand to most fragile category of people such as older individuals, neurogenerative disease affected patients, etc. to lead their normal routine social life being physically active without lowering their quality of life and not losing their autonomy. As ageing progress, the elderly adults experience falls and those consequences lead to lesser quality of life due to the injuries occurred L This further adds to the fear of falling and reduced physical activity. The first solution of protection against the negative effects of falling is to identify fragility and high fall risk at an early stage. The development of novel techniques to identify potential fallers before any fall occurs would be the second solution of protection. The Artificial lntelligence (AD technique is used to examine and prevent falls from the signal-hised image gathered from the gyroscope or the accelerometer sensor information. The several layers of convolutional neural network are used with different dilation rates for extracting different categories of feature selection. The performance evaluation hised on the experimental results on the signal image dataset proves higher accuracy of classification rate.