AUTOMATED SERVER DESIGN AND IMPLEMENTATION FOR SMART SURVEILLANCE SYSTEM

• Main Author: Adams, Kent
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/87693
• Institution: Institut Teknologi Bandung

The improvement of humanity’s quality of life is the basic purpose of any technological invention. With the accelerated technological development rate, surely would bring improvement in humanity’s quality of life. In an attempt to achieve that improvement, the concept of smart city sprang up. In smart city, the very core technologies used are ICT, IoT, Automation, and AI. Smart city itself consist of various sectors, such as healthcare, mobility, energy, waste, community and engagement, economic development, and security. Where, in particularly, security sector has a lot of potential to uncover. One of the implementations of smart city in security sector is in the form of smart surveillance. Smart surveillance has many benefits to offer, such as environmental monitoring, traffic management, public security, and crime prevention. Smart surveillance is built upon a long history, from the very first conception for war purposes, until modern implementation with AI. The rapid pace of development makes smart surveillance a critical advancement, driven by the development of edge devices such as camera, or the processing device such as NVR and VMS. In 2012, there is an idea of removing NVR dependency and use edge storage. But the idea is scratched off in 2014 since reliability of the data and storage capacity became the concern. These considerations push forward the idea of this capstone project. Based on the advantages of smart surveillance in context of smart city, that strive to improve humanity’s quality of life, smart surveillance can be developed into something useful. Evaluating the 2012 attempt, where the utilization of edge device failed, the focus shift onto processing device, such as server. Indeed, within the framework of smart city, the server should utilize smart city technologies including ICT, IoT, Automation, and AI. With those considerations, emerged an idea of “Automated Server Design and Implementation for Smart Surveillance System”. iv In “Automated Server Design and Implementation for Smart Surveillance System”, the idea of automation focused on data processing. Where the scope ranges from stream fetching method until the report received by user. The method used in fetching stream is RTSP protocol, paired with ONVIF standard, compared to Zoneminder software. The fetched stream will be displayed on a website front end using combination of HTML, CSS, and JS via the website back end using Flask. Then, the stream will be processed by detecting motion using frame differencing and gaussian blur algorithm, followed by object detection using YOLO, face detection using Dlib, and face analysis using DeepFace. The analysis result is stored in a report that’s generated by FPDF library and sent to the user via email using smtplib library. After the report is successfully sent, user will receive WhatsApp and Telegram notification, that is implemented using Twilio and Telegram Bot library. This server will run on Ubuntu OS as the base. The implemented system shows promising result, with the whole process ranging from stream processing until report generation on average took around 995.47 ms, which is less than one second. This shows the realistic potential in implementing an automated server in existing smart surveillance system.