Learning-based 3D model creation apparatus and method
AssigneeElectronics and Telecommunications Research Institute
InventorsSeong-Jae Lim, Tae-Joon Kim, Seung-Uk Yoon, Seung-Wook Lee, Bon-Woo Hwang, Jin-Sung Choi
Life of this patent
Estimating the term…
Abstract
Disclosed herein are a learning-based three-dimensional (3D) model creation apparatus and method. A method for operating a learning-based 3D model creation apparatus includes generating multi-view feature images using supervised learning, creating a three-dimensional (3D) mesh model using a point cloud corresponding to the multi-view feature images and a feature image representing internal shape information, generating a texture map by projecting the 3D mesh model into three viewpoint images that are input, and creating a 3D model using the texture map.
Claims13
A method for operating a learning-based three-dimensional (3D) model creation apparatus, comprising:
- generating multi-view feature images using multi-view image generation networks;
- creating a 3D mesh model using the multi-view feature images and internal feature images representing internal shape information;
- generating a texture map by projecting the 3D mesh model into multiple viewpoint images;
- and
- creating a 3D model using the texture map,
- wherein the multi-view image generation networks include a global network corresponding to a global camera group and a local network corresponding to a local camera group.
A learning-based three-dimensional (3D) model creation apparatus, comprising:
- at least one processor;
- and
- a memory for storing at least one instruction executed by the at least one processor,
- wherein the at least one instruction is configured to:
- generate a multi-view feature images using multi-view image generation networks,
- create a 3D mesh model using the multi-view feature images and internal feature images representing internal shape information,
- generate a texture map by projecting the 3D mesh model into multiple viewpoint images, and
- create a 3D model using the texture map,
- wherein the multi-view image generation networks include a global network corresponding to a global camera group and a local network corresponding to a local camera group.
Dependencies are inferred from the claim text.