Recognition of Human Activities in a Controlled Environment using CNN

Human-centered computing is an emerging field of research aiming at understanding human activity and integrating users with computer systems and their social context. Human activity recognition (HAR) can be done using sensors, smartphones or datasets for images/videos. Designing a method in order to understand physical activities such as standing, downstairs, jogging, walking, running, and sitting is very crucial. Convolution neural network (CNN) is used to classify the aforementioned activities with the help of dataset and machine learning. The dataset is used as an input to directly train CNN without any specific pre-treatment. Human activity recognition requires the experimental data so the initial step is to analyze the data after that suitable method is applied to acquire the position of independence. The similarity in activity recognition process was calculated and lastly the study was implemented by compiling experimental data and the effect of different methods. The proposed solution has achieved a high accuracy on identification with low computational costs.