Human Activity Recognition Based on Neural Networks and Accelerators

Le résumé fourni par la source

This study aims to recognise accelerometer-based human activities using ANN models. We used the UCI-HAR dataset, which is a widely used publicly available dataset containing accelerometer data for six daily activities (walking, walking up stairs, walking down stairs, sitting, standing and lying down). In order to improve the accuracy of the recognition, we performed detailed pre-processing of the data, including data cleaning, feature extraction and normalisation. The preprocessed data is fed into the designed ANN model. During the model training process, we used the prediction rate, recall rate and F1 score as performance evaluation metrics. The experimental results show that our ANN model can effectively recognise different activity types, showing high accuracy and robustness. This study not only provides a new approach for human activity recognition, but also provides a valuable reference for the development of intelligent health monitoring systems.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.