HUMAN ACTIVITY RECOGNITION USING ANDROID STUDIO SMARTPHONE
International Research Journal of Modernization in Engineering Technology and Science · 2023
Human activity recognition is the study of human movement and posture through the use of sensors that make predictions about human motion and posture.Using the three-dimensional sensors already present in the average smartphone, this HAR model has been researched and tested.The Accelerometer and Gyroscope are two examples.Input from these sensors is used to make deep learning-based predictions about the activities of moving objects.For performance forecasting, this CNN (Convolutional Neural Network) employs LSTM (Long Term Short Memory).Because of its widespread availability, the proposed system is carried out entirely on a Smartphone (Android Application).After collecting data from sensors and filtering it, we train TensorFlow on the data using the prepared dataset.We also compare the accuracy of CNN models using various optimizers and explore the implications.To show how accurate the model is, we built an app for Android that can do the job.We think that after some tweaks, this software might be used by the elderly who live alone to report emergencies to the proper authorities in a timely manner.