ARC Synthetic Riverine Waste

Abstract

The Autonomous River Cleanup (ARC), led by students at the Robotic Systems Lab, aims to tackle waste pollution in rivers using robotics and machine learning. The team consists mainly of volunteers and students, with guidance from experienced researchers, and focuses on waste analysis and sorting. This dataset results from a synthetic waste generation pipeline developed by Xenia Voellmy and Dario Strübin as part of their Master's thesis at ARC. Real waste items were digitized in 3D, then deformed and textured using Blender to simulate the variations commonly found in riverine waste. A set of RGB images, depth images, and segmentation masks are provided containing multiple synthetic waste items on randomized backgrounds with item-wise annotations in the coco format.