Install what we build.

One vision model for bounding boxes, polygons, polylines, keypoints and tags — train them together in a single model or one label type on its own, on datasets of any size.