Inspiration
Originally this was going to be a project about tracking crowds using a multi-node IR system. We maintained the multi-node aspect of it, and transitioned to trying to track a single person.
How it works
OpenCV is used for blob tracking on two Lepton cameras. If all goes according to plan, one node will tell the other node to start searching for a person.
Challenges I ran into
Getting the image processing to work was very challenging because there are a lot of weird data type incompatibilities and scaling issues. We also hit a lot of bugs at the last minute with the network cutting out on us and such.
Accomplishments that I'm proud of
If the robot is able to follow a person's heat signature that is a huge accomplishment for us.
What I learned
We learned a lot about image processing, about GPIO on the Raspberry Pi, and about running socket servers in node.js and in Python.
What's next for WatchOn
More nodes and crowd tracking for larger events. It seems like the multi node system doesn't really work right now, oh well.
Log in or sign up for Devpost to join the conversation.