CSGNet: Neural Shape Parser for Constructive Solid Geometry
share
We present a neural architecture that takes as input a 2D or 3D shape and induces a program to generate it. The in- structions in our program are based on constructive solid geometry principles, i.e., a set of boolean operations on shape primitives defined recursively. Bottom-up techniques for this task that rely on primitive detection are inherently slow since the search space over possible primitive combi- nations is large. In contrast, our model uses a recurrent neural network conditioned on the input shape to produce a sequence of instructions in a top-down manner and is sig- nificantly faster. It is also more effective as a shape detec- tor than existing state-of-the-art detection techniques. We also demonstrate that our network can be trained on novel datasets without ground-truth program annotations through policy gradient techniques.
READ FULL TEXT