A few months ago, a van drove into the sidewalk of Toronto and killed 10 people and injured 15. We wanted to find solution to help prevent and reduce the potential casualties when incidents like this occur so we decided to create IoT system that would aid the people who are most in need of help.
What it does
We use computer vision technology to detect the presence of cars driving into a sidewalk. When detected, an audible alarm will trigger to alert people around the area so that they can hopefully avoid/get away from the vehicle. A SMS message will also trigger to notify people in the area that there is a potential danger on the sidewalks and to remain inside buildings to stay safe until cleared.
How we built it
We used OpenCV and Tensorflow to classify objects in a video stream. When a Car was identified, it would send a real time message to a buzzer and to a local computer that would send an SMS message with stdlib's SMS function.
Challenges we ran into
When initially developing the MVP, we thought we could use boundaries to detect when a car was on a sidewalk. This proved to be a bit more challenging than anticipated so we decided to change the requirements to simply detecting a car in the video stream as the camera can always just point at only the sidewalk as a prototype.
Accomplishments that we're proud of
We completed the MVP and also added additional features like a front-end for local authorities to review for faster reaction time and a way to control the alarm systems.
What we learned
OpenCV is not a simple program to learn especially when your developers are not familiar with the programming enviroment (Python) used. We also learned how to better design tasks so that they are less intimidating to the developers.
What's next for SSS
The idea would be to use this technology to expand into illegal parking, shootings, asaults and other criminal offenses.
HackThe6IX - SSS Smart Surveillance System -> Oscar Kwan, Tancred Yip, Timothy Wang
See [LICENSE] for details.
Copyright (c) 2018 [Oscar, Tancred, Timothy]