Advanced IoT-Based Activity Recognition System Using Meerkat Optimization Algorithm with Feature Selection Model for Elderly and Disabled People

In essence, the Internet of Things (IoT) is a new technology that is suitable for numerous systems and devices that are applied every day, like cellular phones, computers, sensors, with other related devices and people. Nowadays, the IoT plays an important role in guaranteeing that the present quality of life is more comfortable, easier, and better than it was before. Recently, constantly growing technological developments have made automatic human activity recognition an ordinary area of research. Video surveillance has a wide range of applications. These applications consist of suspicious and normal actions like analyzing sports, gaming, exam invigilation, human-computer interaction, predicting crowd behavior, detecting chaos, and so on. It is a significant safety feature for outdoor and indoor atmospheres. The activity detection of humans depends on sensors to recognize anomalies in human behavior. Currently, Deep learning (DL) is the most typical framework that learns challenging tasks amongst other designs. This paper presents an Activity Recognition System Using Meerkat Optimization Algorithm with Feature Selection Model (ARSMOA–FSM) in IoT for Disability People. The main goal of the ARSMOA–FSM technique is to deliver an effective IoT-based activity recognition model for disabled people using advanced techniques. Initially, the data pre-processing stage applies Z-score normalization to convert the input data into a structured format. For the selection of feature process, the Arithmetic Optimization Algorithm (AOA) has been deployed to select the most relevant and significant features from the data. Furthermore, the proposed ARSMOA–FSM model designs a bidirectional gated recurrent unit (BiGRU) model for the classification process. At last, the parameter tuning process is performed through the Meerkat Optimization Algorithm (MOA) to improve the classification performance of BiGRU classifier. The experimental evaluation of the ARSMOA–FSM model occurs using a benchmark dataset. The stimulated results indicated the enhanced performance of the ARSMOA–FSM model compared to recent approaches.