BenDFM: A Synthetic Dataset for Manufacturability Assessment in Sheet Metal Bending

This repository hosts BenDFM, a large-scale synthetic dataset of sheet metal parts designed for data-driven Design for Manufacturing (DFM) research. The dataset is introduced in the paper: BenDFM: A taxonomy and synthetic dataset for manufacturability assessment in sheet metal bending and provides: Geometrically diverse 3D sheet metal bending parts and their unfolded representations. Rich part-level metadata. Comprehensive manufacturability labels structured according to a novel taxonomy, spanning both geometric and configurational feasibility and complexity. BenDFM enables a wide range of tasks, including binary feasibility classification, regression of manufacturing complexity, and analysis of process constraints. The dataset is split into two subsets: BenDFM and BenDFM-U, each balanced to support specific manufacturability tasks. For more information, we refer to the dedicated Github repository: BenDFM