| Autor: |
Iswarawati, N. K. E., Satyawan, A. S., Puspita, H., Utomo, P. A., Putri, R. A. |
| Předmět: |
|
| Zdroj: |
Internet of Things & Artificial Intelligence Journal (IOTA); May2024, Vol. 4 Issue 2, p187-198, 12p |
| Abstrakt: |
The airport is a strategic place where it involves important activities, namely airplane flights. Airport activities are greatly influenced by the ongoing security within the airport. The thing that affects security in the airport is the possibility of crimes committed by unexpected visitors. The purpose of this research is to observe the airport area, especially airport visitors so that if there are visitors who have the potential to commit crimes, they can be detected properly and further investigation procedures can be carried out. To be able to observe airport visitors and recognize patterns of visitor behavior that have the potential to commit crimes, an airport visitor gesture recognition system can be used. In this thesis, the gesture recognition of airport visitors is done with the multipose estimation method. This method can detect 17 key points on the human body that are used to detect the behavior of airport visitors who have the potential to commit crimes. To develop this system, deep learning algorithms that are currently developing with the help of TensorFlow and architectural models in multipose estimation, namely MobileNetV2, Feature Pyramid Network, and CenterNet, can be used. The experimental results show that the multipose estimation method can recognize human gestures well under several conditions such as the appropriate distance of the human object from the camera and the lighting conditions around the observed human object. It is also seen that from several scenarios, the crime gesture model can be recognized well. [ABSTRACT FROM AUTHOR] |
| Databáze: |
Complementary Index |
| Externí odkaz: |
|
