Abandoned Object Detection Method Using Convolutional Neural Network

Abstract

Automatic surveillance is an effort to detect anomalies that occur in the surrounding environment such as stations, offices and other public spaces. One of the anomalies that occurs is neglected objects. Abandoned objects will become annoying or dangerous if left unattended. Abandoned object detection process begins by detecting a stationary object using Gaussian mixture models, then abandoned recognize objects using convolutional neural network. The recognition of stage objects is very helpful in determining the bounding box of objects that are left behind, thereby reducing the bias arising from shadows or lighting changes. The resulting effective method to detect abandoned objects and recognize it.