This document presents a garment-agnostic approach for robotic garment unfolding. The proposed approach uses a 3D mesh reconstructed from an RGB-D sensor to segment and cluster the garment. Regions are selected for unfolding based on bumpiness. Experiments using an industrial robot on 6 garment categories showed a 20% performance improvement over the previous approach. Future work includes improving segmentation of thin garments and experimenting with tactile feedback during manipulation.