Human Activity Recognition and Prediction using LSTM and Transformer

Human Activity Recognition (HAR) in AI plays a crucial role in enhancing the lives of individuals by providing health monitoring, safety, and assistance. The proposed work focuses on recognition and prediction of the Human activity by combining the concepts of LSTM and Transformer, enabling HAR systems to identify sudden falls or unusual inactivity, automatically alerting caregivers or emergency services. The performance shown by the combination of the methods LSTM and Transformer loss delivers improved performance compared to existing approaches such as Hidden Markov models and LSTM.