Sensor Based Human Activity Recognition by Multi-Headed ConvLSTM
Dheeraj Shah, Anushka Malhotra, Madhur Patidar · 2021
The past few years have witnessed a boom in the field of Artificial Intelligence, driving people towards it. Various sectors around the globe are using artificial intelligence, especially the Health-care sector. The healthcare sector today has started using sensor based technology in order to automate the activity of people belonging to any age group. As the name suggests, the sensor based method can be elaborated as an approach to record or “sense” activities of various age groups, and sensor data is being used for the same. Earlier, this was carried out using the machine learning models, but as the techniques progressed, researchers became more inclined towards the latest techniques rather than the traditional ones, the most recent one being Deep Learning. This particular paper proceeds in that direction as well. The basic idea behind this paper is to bring forth a new technique for Human Activity Recognition (HAR). The technique used here is “MultiHeaded ConvLSTM”, to build a deep neural architecture for improved accuracy of HAR.