This repo is for experiments using Raster Vision to measure the performance of algorithms under a variety of different circumstances. The repo is public for the sake of repeatability and transparency, but doesn't aim to be very well-maintained, supported, or documented. In contrast, the raster-vision-examples repo contains examples of using Raster Vision on open datasets, and is a designed to be a good starting point for beginners.
- noisy-buildings-semseg: How does error in training labels affect accuracy for semantic segmentation?