This is a helper program written for the server-side of Xorzee, which is largely written under node-js. Due to performance needs, this is written in Rust.
The program expects a stream of 32-bit motion vectors fetched while the GPU is encoding the sensor data to H264, making it essentially free. This program will then output a stream of JSON messages passed to the node-server. Node-js will process it for activity and then pass it on over websockets to clients.
It operates in the following ways on these motion vectors:
- Parse
- Filter
- Categorize
- Density based cluster
- Temporal tracking
- Merge
I think the density based scan is quite fast. That said, the way things
are implemented, epsilon might not mean exactly what you would expect.
But as long as I am not using euclidean distance, it matters little.
Stand-alone, without Xorzee, this is probably of little use to you. :-)
You need to install arm linker; arm-linux-gnueabihf-gcc. This section should probably be fleshed out.
cargo build --release --target armv7-unknown-linux-gnueabihf
- I want to get rid of JSON and use a more binary format. Not because it's terrible in here, but because it takes a few milliseconds to parse it on the Node side if there is a lot of activity.
- Make configurable and finalize command-line arguments