Bionote
Gianmarco Roggiolani is a Ph.D. candidate in the Photogrammetry & Robotics Lab at the University of Bonn, Germany. He obtained his B.Sc. degree in Computer and Automatic Engineering in 2018 and received his MSc degree in Artificial Intelligence and Robotics in 2021, both from the Sapienza University of Rome, Italy. His research focuses on self-supervised techniques to improve the performance of vision-based learning systems in agricultural robotics reducing the labeling effort.
Presentation Abstract
Robust perception systems allow farm robots to recognize weeds and vegetation, enabling the selective application of fertilizers and herbicides. Today’s perception systems typically rely on deep learning to interpret sensor data, but they require substantial amounts of manually labeled training data, which is often time-consuming and requires domain expertise. In this presentation, we show how to reduce this limitation with an automated labeling pipeline for crop-weed semantic image segmentation. Our system exploits the field row structure for spatially consistent labeling and evidential deep learning to refine and improve our predictions.
Co-Authors: Julius Rückin, Marija Popovic, Jens Behley, and Cyrill Stachniss