This lightweight web API is intended to run alongside the Smart Bin Server, allowing computer vision inference to be offloaded to a separate GPU node, while our database and public web API for serving compost bin clients runs on a storage node.
This API exposes one route, which allows a client to send an image to the server, and returns an image with segmentations drawn overtop, as well as a results.txt which provides labels and polygon coordinates for each detected food waste item.
We use YOLOv8 for our image recognition model. Training code for this model can be found here