Elderly fall detection using IoT and image processing. (3rd April 2021)

• Record Type: Journal Article
• Authors: Kumar, Vinod; Badal, Neelendra; Mishra, Rajesh
• Is Part Of: Journal of discrete mathematical sciences & cryptography. Volume 24:Number 3(2021)
• Page Start: 681
• Page End: 695
• Publication Date: 2021-04-03
• Subjects: Internet of Things; Net acceleration -- Pitch -- Roll; Computer science -- Mathematics -- Periodicals; Cryptography -- Periodicals; Computer science -- Mathematics; Cryptography; Periodicals; 004.0151
• DOI: 10.1080/09720529.2019.1692451
• Languages: English
• ISSNs: 0972-0529

Abstract

Abstract: The falls due to accidents or loss of consciousness in older people can turn out to be fatal if the required assistance is not received with in time. The fear of falling puts constraints on the free movement of the elderly and forces them to always be accompanied by a caretaker. This paper deals with the problem to seek help as and when required rather than always being watched over. The key contribution of this paper belongs to its novelty to detect elderly fall and even to cross validate it(that increase the trust for the model). To find the fall, algorithm fallows 2-way approach. In first way the sensors collect all the data related to fall such as Net acceleration of body, Pitch and Roll, based upon these values various activity is identified. Secondly, to further increases the trust over the model (remove false positive scenarios), this paper has also used image processing. Here, the images of various finger sign made by elderly person are processed at the time of fall. Thus, our work in this paper (Algorithm for fall detection using IoT and Image Processing(AFDI2) works over all the possible conditions and gives 99% accuracy mostly in all the possible conditions whereas SVM model only able to reach an accuracy of 97%, 96%, 98% while walking, walking downstairs and walking upstairs.