Human Activity Recognition System using Accelerometer and Gyroscope Data

Using reactive sensors of smartphones that are affected by human activity, human activity recognition coordinates an individual's activity. In the presented paper, the proposed model used a combination of Convolution neural networks (CNN) and Long Short-Term Memory (LSTM) network for recognition of human activities from the accelerometer and gyroscope data. The proposed model consists of three separate CNN. Each CNN is designed to extract features from the data. Feature extraction is followed by an LSTM layer, which is a type of recurrent neural network (RNN), for making predictions based on the features extracted. This method delivers high accuracy. The performance metrics is shown in this paper in detail.