Smart Victims Detection in Natural Disaster using Deep Learning

Nowadays, artificial intelligence has been widely implemented in agriculture, animal husbandry, education, and many more. The most widely developed model at this time is the Deep Learning model. To get maximum results, the best Deep Learning architecture is needed. In this paper, we will compare methods for detecting victims of natural disasters based on the Deep Learning model. The models that we will compare are the You Only Look Once (YOLO) algorithm, namely YOLOv5, and Convolutional Neural Network (CNN) with the MobileNet model and the VGG-16 model. We train each algorithm through a natural disaster victim training data set and analyze performance to determine what model is the most optimal. To identify the types of victims of natural disasters. From our results, the MobileNet model is superior to other models with 98% accuracy.