An international conference connecting people
in CAD research, education and business
Bookmark and Share
Copyright (C) CAD Solutions, LLC. All rights reserved.
Proceedings of CAD'16, 2016, 200-204
Introduction of Learning Techniques for Creating Solid Models from Sketches Including Curves

Masaji Tanaka, Yuki Takamiya, Naoki Tsubota, Shigeo Asanuma, Okayama University of Science
Kenzo Iwama, EngiCom Corporation

Abstract. Since several years, the authors have developed a method to automatically convert sketches as 2D line drawings including curved lines into 3D models. However, it was difficult and time consuming to develop the experimental system of the method because various kinds of geometric pattern matching processes had to be programmed. On the other hand, since decades, machine learning techniques have been developing very quickly. Since the techniques are effective for pattern matching, we apply our inductive learning techniques to the method in our new method proposed in this paper. There are many machine learning systems. Especially, deep learning has become popular and has its applications in pattern matching. However, it is not clear how a deep learning method constructs a procedure out of example sequences of steps unless the method produces each step out of example steps and some other method puts the steps into a sequence of the steps. The learning technique of our method could construct a procedure from example sequences of steps. For example, our method can learn a procedure to calculate x-y coordinates of the intersection of two straight lines by generalizing a few examples. Moreover, the procedure can be applied as a function to the other procedures, e.g. a procedure that can divide two lines at their intersection.

Keywords. Inductive Learning, IFOG, Sketch, Solid Model

DOI: 10.14733/cadconfP.2016.200-204