Registered domain name with domain.com: raccoongaze.online
Chad wanted a setup to watch over his baby raccoons.
What it does
A web app where you can watch the live-stream of the baby raccoons, identify them, do live object tracking and other cool stuff, all in real-time.
How we built it
We set up a live-stream using RTSP ( Real Time Streaming Protocol ) and streamed it using a Raspberry Pi from sharing out the LAN into a switch and then we setup a Ubiquiti Unifi AP to get WiFi to the habitat.
We set up a docker container that handled the live-stream performed Real Time Analysis on it using the Google Cloud Platform. (The Video Intelligence Streaming API supports standard live streaming protocols like RTSP, RTMP) and we then deployed it on Kubernetes.
Now it acted like a proxy for the stream (The AIStreamer ingestion pipeline behaves as a streaming proxy, converting from live streaming protocols to bidirectional streaming gRPC connection).
The output stream along with the labeling and object tracking data was then set up for viewing through a React web application ( with a few easter eggs).
Challenges we ran into
We couldn't get the RTSP set up, we tried various hardware and software fixes.
The Video Intelligence API is in Beta, so documentation was scarce and we faced a lot of hurdles.
Accomplishments that we're proud of
We were able to work remotely across different time-zones, and we were able to pull something off in this short amount of time. As well, we incorporated many technologies in our stack, all while overcoming sleep.
What we learned
Setting up a network for streaming, and using real-time video feeds from the raccoon home. As well, we learned how to incorporate something laid back and cute (raccoons) with an extensive project. Finally, we learned how to tie in the Raspberry Pi video feed to an UI for the user over AngelBox.
What's next for RacoonGaze
Distribution to new mothers, new pet owners, and anybody that wants to keep track of animals, people, or any object!