Overview / Usage

The goal of this project is to improve the ability to respond to flooding in cities by leveraging the real-time video streams from traffic cameras. The video streams will be analyzed using the analytics pipeline available within the Open Visual Cloud toolset.

Methodology / Approach

I decided to leverage the existing sample in Github.( https://github.com/OpenVisualCloud/Smart-City-Sample). I explored the Intel DevCloud first as a hosting platform but ran into issues related to installation of some the packages. Then I decided to do an installation on a Dell PowerEdge Xeon Server that was running Ubuntu Server 18.04. I ran into several challenges to get the sample running and created issues in Github to alert the repo owners. After hacking through them, I was eventually able to get the website up. I replaced one of the existing video files in the sample with a publicly available road flooding video. Clicking that camera in the map now shows the flooding video. The leftmost lane is flooded and you can water splashing as cars drive through flooded road. I looked into adding analytics to detect a flooded road but ran into issued because the analytics pipeline seems to malfunction on a single server docker swarm installation. I have documented the issue in github. It is possible that the pipeline works in a multi-node swarm. One the OpenVisualCloud Github repo is updated with a fix, I am planning to redeploy and explore the analytics engine. I am also planning to see how the Neural Compute Stick 2 can integrated to offer cost-effective real-time compute power. Once the prototype is functional, the next step would be to do a field-test with a large city such Seattle. (https://web6.seattle.gov/travelers/) that has a large camera network open to the public. A larger goal is to apply the framework to a variety of first response situations including active shooter and fires. I am part of the team that has built an IoT Data generation appliance for another hackathon (https://www.challenge.gov/challenge/chariot/#description) and we are one of the two finalists (the final judging will be later this year). My goal is to integrate the data from our Smart IoT Appliance into this POC to leverage sensor fusion.

Technologies Used

Intel Open Visual Cloud, Docker Swarm, Python, Ubuntu 18.04 Server, Dell Xeon PowerEdge Server

Other links

• Seattle Traffic Cameras
• Chariot NIST IoT Appliance Challenge

• Open Visual Cloud Smart City Sample (https://github.com/OpenVisualCloud/Smart-City-Sample)