3D Modelling Using Machine Learning Technique

• Main Author: Zhao, Haolong
• Other Authors: Huang Guangbin
• Format: Final Year Project
• Language: English
• Published: 2018
• Subjects: DRNTU::Engineering::Computer science and engineering::Computing methodologies::Image processing and computer vision
• Online Access: http://hdl.handle.net/10356/75556
• Institution: Nanyang Technological University

The objective of this project is to perform 3D modeling using machine learning techniques, extensive research on 3D modeling and machine learning techniques were conducted. Machine learning methods are classified as the image rending-based methods, it has the features of low cost, flexible in application, easy to set up, which are desired in most of the application scenarios. In-depth study and testing of 3D-R2N2 network is also carried out. LSTM, CNN networks are studied during the process of understand the network structure of 3D-R2N2. As well as dataset preparations including image rendering and voxel grid, which are fundamental steps of machine learning works. Test Results of two dataset, ShapeNet used in previous, and ModelNet40, extra dataset rendered in this project are shown and discussed in the report, too. Basically, 3D reconstruction faces many challenges like self-occlusion, tilted viewing angle, those intrinsic obstacles makes 3D reconstruction using machine learning a very challenging.