Pro-Safe: An IoT based Smart Application for Emergency Help
Emmanuel Abraham, C K Nameer, Roshan Tom, Yedul Ganesh, S S Vidhya · 2019
There can be any sudden situation of panic. It could be because of an intruder entering our house, an accident situation, bad health status or an assault at which we are unable to intimate to the people around us. In this paper, we proposes an implementation of emergency handling system, which is based on fall detection. The proposed system consist of an android based smart device embedded with real time fall detection.The main objective of this project is to build an automated emergency handling device with an auto triggered emergency button. Once the application is triggered it will initiate live recording of images and audio of the situation.The encrypted copy of this will be uploaded in a cloud storage which is accessible by an authentic police official. This application will also give an alert to the nearest police station with the users current location.In this paper, we compare three machine learning algorithms, Decision Tree(DT) algorithm, K-Nearest Neighbour(KNN) algorithm and Support Vector Machine(SVM) algorithm using Sis Fall dataset for training and validation so as to provide high efficiency in fall detection.In this scenario, SVM algorithm gives better accuracy and less false positive rate.Here we used triaxial accelerometer and gyroscope to collect the data. The proposed device achieves high performance by using Machine learning techniques to detect harmful fall of the person and to auto initiate by itself.