Overview :

We propose a solution aimed at making a system which would use surveillance cameras, a microcomputer, and deep learning algorithms to ensure more safety of a place by detecting any suspicious activity happening and report it to the concerned authorities (police etc.) with the location and video stream, if flagged suspicious.
The system proposed is capable of classifying a video stream into :

1. Criminal or violent activity
2. Potentially suspicious
3. Safe.

It is also capable of detecting potential mishaps such as road accidents( nearest hospital will automatically be intimidated), fire accidents (will call for immediate fire fight) etc. Not only this, the proposed system is capable of predicting mishaps before them even happening, based on a probablistic value, and aims to reduce the deaths and accidents happening on roads and highways due to negligence or unavoidable circumstances even before they occur.
So to sum up, our system comprises of two parts:

• Video Analysis to detect suspicious activity
• Road Accidents prevention by predicting possible mishaps beforehand