Design of Smart Door Closer System with Image Classification over WLAN

Abstract

A dual purpose system is presented in this paper which serves not only as a door closer, but is equally effective for surveillance purposes. The currently deployed surveillance systems store a large amount of data, thereby consuming large memory spaces. The novel feature illustrated in this paper is that object identification and classification is performed for a desired area, along with controlled access. This system uses a neural network based learning algorithm before providing any instruction for the hardware. The system is innocuous except when object is identified in the surrounding area through motion detection and facial recognition techniques, thereby preventing large storage of unclassified video frames. In idle conditions, the system will work only as a surveillance system with object detection classifiers. The results can be stored on remote server for backup for security purpose.