| Title |
Detection of Emergency Disaster using Human Action Recognition based on LSTM Model |
| Authors |
(Yull Kyu Han) ; (Young Bok Choi) |
| DOI |
https://doi.org/10.5573/IEIESPC.2020.9.3.177 |
| Keywords |
Emergency detection; Deep learning; LSTM; Human action recognition |
| Abstract |
We propose a deep learning model for human action recognition in order to identify quickly the location and occurrence of disasters such as fire and terrorism. Using the acceleration and gyroscope sensors built in the smartphone, we obtained data on human behavior and we classified human behavior through the LSTM deep learning model. There are four categories of human behavior, usually stopping, walking and running, and running helter-skelter in the event of a disaster. We compared analysis of four types of human behavior data such as stop, walking, running, and running helter-skelter, with the existing NN model and the LSTM model proposed in this paper. As a result, we confirmed that the LSTM model can be classified more accurately than the NN model with 81.50% for the conventional NN model and 95.32% for the LSTM model. We expect that the proposed model can be used to detect disaster occurrences quickly. |