Computer Vision Based Surveillance System

Abstract

Nowadays CCTV cameras are almost everywhere in Bangladesh and the number is increasing rapidly. This CCTV footage is still manually observed and it is too hard for a human to keep eyes on every single camera continuously which is boring and most of the time, because of human weaknesses, could be missed. As a result, various types of incidents occur, which could have been prevented by taking necessary steps in time. During such various movements a large number of cars, shops, government, and private properties are destroyed, maybe by setting things on fire or breaking things that caused great loss to the country. Or people in danger do not get the help they need in time. These days CCTV's and the internet are very cheap. These being used everywhere, CCTV's can be found in every corner of the road, school, college, university, and markets. But there is a lack of an efficient system to use this CCTV smartly. As a result, storage is wasted because they are recording all the time. But a smart system can be turned on recording automatically just whenever it needs to record. And regular CCTV systems do not send notifications or alarms in a proper way. So, most of the time crucial events are missed just because this footage is being checked manually. So, the main purpose of using CCTV being failed. People does not get help at right time and fail to secure their properties. So, we developed a video surveillance system (web application) that will detect fire, hand gestures, and suspicious or unwanted human beings quickly and will inform (Email, SMS and Web notifications) the authorities about these unwanted incidents with proper evidence (photos and videos). It will be possible to save from the huge loss of life and property. And people can get the necessary help urgently.