Faculty Publications
Permanent URI for this communityhttps://idr.nitk.ac.in/handle/123456789/18736
Publications by NITK Faculty
Browse
Search Results
Item Human Activity Behavioural Pattern Recognition in Smart Home with Long-Hour Data Collection(Springer, 2023) Kolkar, R.; Geetha, V.The research on human activity recognition has provided novel solutions to many applications like health care, sports, and user profiling. Considering the complex nature of human activities, it is still challenging even after effective and efficient sensors are available. The existing works on human activity recognition using smartphone sensors focus on recognizing basic human activities like sitting, sleeping, standing, stair up and down, and running. However, more than these basic activities is needed to analyse human behavioural pattern. The proposed framework recognizes basic human activities using deep learning models. Also, ambient sensors like PIR, pressure sensors, and smartphone-based sensors like accelerometers and gyroscopes are combined to make it hybrid sensor-based human activity recognition. The hybrid approach helped derive more activities than the basic ones, which also helped derive human activity patterns or user profiling. User profiling provides sufficient information to identify daily living activity patterns and predict whether any anomaly exists. The framework provides the base for applications such as elderly monitoring when they are alone at home. The GRU model’s accuracy 95% is observed to recognize the basic activities. Finally, Human activity patterns over time are recognized based on the duration and frequency of the activities. It is observed that human activity pattern, like morning walking duration, varies depending on the day of the week. © 2023, The Author(s), under exclusive licence to Springer Nature Singapore Pte Ltd.Item Human activity recognition using deep learning techniques with spider monkey optimization(Springer, 2023) Kolkar, R.; V, G.The human activity recognition (HAR) system recognizes human actions in daily life. There is a need for HAR to build a smart home and an intelligent healthcare environment. HAR is challenging, considering the complexity and heterogeneity of sensors used to recognize it. Deep learning models are the one area where the researcher applies to recognize the activities. However, effective feature engineering and optimization methods help improve the recognition model’s performance. In this work, Spider Monkey Optimization is applied for training the deep neural network. UCI HAR, WISDM, KTH action and PAMAP2 datasets are used to evaluate the proposed system. The dataset has the activities like walking, standing, lying, jogging, stair-up and stair-down activities. Here, the spider monkey model’s fitness function is initialized in the hidden layer of the Recurrent Neural Network to enhance accuracy and precision. The experiment results show improvements in performance as compared to other state-of-the-art methods like DL-Q, End to End DNN and SVM. With various assessments and experimentation, it is observed that the proposed SMO-based performs better in terms of accuracy of 98.92%, precision of 98.12%, recall of 98.9%, and F1-score 95.90%, respectively for the WISDM dataset. There is an improvement in performances for other datasets. Also, the Error rate has reduced to 2.8% as compared to other state-of-the-art methods. © 2023, The Author(s), under exclusive licence to Springer Science+Business Media, LLC, part of Springer Nature.