[Human activity recognition based on the inertial information and convolutional neural network].

Area of Science:

• Computer Science
• Biomedical Engineering
• Wearable Technology

Background:

• Mobile devices offer a convenient platform for human activity recognition.
• Inertial data from smartphone accelerometers provide insights into human motion.
• Traditional computer vision methods have limitations in cost and convenience for activity recognition.

Purpose of the Study:

• To develop and evaluate a human activity recognition model using deep learning on smartphone inertial data.
• To compare the performance of a convolutional neural network (CNN) against traditional machine learning algorithms (KNN, Random Forest).

Main Methods:

• Utilized the WISDM dataset collected from smartphones.
• Employed inertial navigation data from smartphone acceleration sensors.
• Developed a human activity recognition model using a convolutional neural network (CNN).
• Compared CNN performance against K-Nearest Neighbors (KNN) and Random Forest algorithms.

Main Results:

• The CNN model achieved a classification accuracy of 92.73%.
• CNN significantly outperformed both KNN and Random Forest algorithms in recognition accuracy.
• The results validate the effectiveness of deep learning for human activity recognition.

Conclusions:

• Convolutional neural networks provide a highly accurate method for human activity recognition using smartphone sensors.
• This approach has significant potential for applications in health prediction and promotion.
• The study highlights the advantages of using inertial data and deep learning for unobtrusive health monitoring.