Disaster Victims Detection System Using Convolutional Neural Network (CNN) Method

Abstract

Natural disasters are one of the things that cannot be predicted. Natural disasters can cause losses, both assets and objects can even take lives. To reduce the number of losses, rapid evacuation handling from the Search and Rescue (SAR) team is needed to help victims of natural disasters. But in fact, there are often obstacles in the evacuation process. Such obstacles are such as bad weather conditions, disconnection of telecommunications networks, difficulty access to the victims of natural disasters and the spread of SAR teams that are not evenly distributed throughout the disaster area. Convolutional Neural Network is one of the developments of Artificial Neural Networks for image classification, image segmentation, and object recognition with high accuracy and high performance. CNN can learn to detect various images according to images from the dataset studied. So, this paper designed a system for detecting victims of natural disasters using the CNN method and implemented it on a raspberry pi which can detect victims of natural disasters through streaming cameras placed on UAVs. In this paper, the Convolutional Neural Network (CNN) method with 100% accuracy with distance object 1–4 m uses the Mobile-net SSD model.