I am thrilled to share our new paper SCENE-Net: Geometric induction for interpretable and low-resource 3D pole detection with Group-Equivariant Non-Expansive Operators is now available in Computer Vision and Image Understanding 🎉🎉.
In this paper, we introduce our ML model called SCENE-Net for the identification of towers from 3D point cloud data. We present the model that was built using GENEOs, explain its training procedure, and evaluate it from the perspective of various metrics. Most importantly, SCENE-Net needs minimal resources compared to other state-of-the-art models, marking a significant step forward in trustworthy machine learning applied to 3D scene understanding