Is Fusion 360 Reconstruction Dataset Really Not Enough for Training When Compared With DeepCAD Dataset?

This article seriously investigates the two typical datasets in computer-aided design (CAD) sequence learning. The DeepCAD (with 178 k samples) paper indicates that the limited size of the Fusion 360 dataset (less than 10 k samples) is insufficient for training a well-generalized model. Thus, almost all of the top conference and journal papers adopt the DeepCAD dataset for training. However, this investigation reveals that, although there is a huge gap in data size between the Fusion 360 Reconstruction and DeepCAD datasets, they have almost the same ability in CAD sequence learning. We devise reasonable experiments and a data augmentation method to demonstrate that the Fusion 360 Reconstruction and DeepCAD datasets are essentially indistinguishable, exhibiting equivalent capabilities in CAD sequence learning for simple sketch and extrusion commands. Therefore, to advance the development of CAD sequence learning, we need more complex and advanced CAD datasets, which is a more challenging task for our community in the future.