semantic-segmentation-editor

Why we included this project

This web tool covers both halves of the labeling problem: you can draw semantic segmentation polygons on JPGs and PNGs, then switch to PCD point clouds, including binary compressed and RGB-colored scans, all in one interface. It grew out of autonomous driving research, so it has seen real use on sensor data. The app is built with Meteor and React and ships with a Docker Compose stack, which means a small team can run it on their own hardware without paying per-seat or sending proprietary data to a cloud service. If you need a self-contained labeling setup that handles images and lidar-style point clouds without a commercial platform, this is a reasonable starting point.