Method and apparatus for generating virtual object

1. A method of generating a virtual object, the method comprising:

acquiring a point cloud that is generated based on a depth map of an object;

determining shape attribute information of the object based on an image of the object using a neural network trained to recognize a shape of the object by generating the shape attribute information as respective information with respect to a plurality of shape attributes comprising at least two of a mirror symmetry, a rotational symmetry, a thin structure, a rough surface, or a smooth surface, wherein an element, corresponding to a vector having the respective information as respective elements of the vector, is a binary value;

changing a position of at least one point in the point cloud, based on the shape attribute information, comprising;

changing a position of a first point in the point cloud, to correspond to a first element of the vector; and

changing a position of a second point in the point cloud, to correspond to a second element of the vector;

and

generating a virtual object for the object based on the point cloud in which the position of the at least one point is changed.

2. The method of claim 1 , wherein the acquiring of the point cloud comprises:

generating a depth map based on the image; and

generating the point cloud based on the depth map.

3. The method of claim 2 , wherein the generating of the depth map comprises generating depth maps for multiple viewpoints of the object.

4. The method of claim 1 , wherein the acquiring of the point cloud comprises:

generating the depth map of the object using a depth camera; and

generating the point cloud based on the generated depth map.

5. The method of claim 1 , wherein the generating of the virtual object comprises:

generating a three-dimensional (3D) mesh based on the point cloud in which the position of the at least one point is changed; and

generating a texture of the 3D mesh.

6. The method of claim 5 , wherein the generating of the texture of the 3D mesh comprises generating the texture of the 3D mesh, in response to the image being a color image.

7. The method of claim 1 , wherein the virtual object is included in any one or any combination of augmented reality (AR) content or virtual reality (VR) content.

8. The method of claim 1 , wherein a number of times the position of the at least one point in the point cloud is changed corresponds to a number of elements of a vector having the respective information as respective elements of the vector.

9. The method of claim 1 , wherein the determining of the shape attribute information includes applying image information of the image to a machine learning trained multi-attribute classification model that outputs respective information that indicate which of plural attributes in the image are resultatively classified as being attributes of the object.

10. A non-transitory computer-readable storage medium storing instructions that, when executed by a processor, cause the processor to perform the method of claim 1 .

11. An apparatus for generating a virtual object, the apparatus comprising:

a processor; and

a memory storing instructions that when executed by the processor configure the processor to:

acquire a point cloud that is generated based on a depth map of an object;

determine shape attribute information of the object based on an image of the object using a neural network trained to recognize a shape of the object by generating the shape attribute information as respective information with respect to a plurality of shape attributes comprising at least two of a mirror symmetry, a rotational symmetry, a thin structure, a rough surface, or a smooth surface;

change a position of at least one point in the point cloud based on the shape attribute information;

change a position of a first point of the at least one point in the point cloud, to correspond to a first element of the vector;

change a position of a second point of the at least one point in the point cloud, to correspond to a second element of the vector; and

generate a virtual object for the object based on the point cloud in which the position of the at least one point is changed.

12. The apparatus of claim 11 , wherein the processor is further configured to:

generate a depth map based on the image; and

acquire the point cloud based on the generated depth map.

13. The apparatus of claim 11 , further comprising a memory configured to store any one or any combination of the depth map of the object, the image of the object, and the virtual object.

14. The apparatus of claim 11 , wherein the neural network is a trained multi-attribute classification model that outputs a vector, where respective elements of the vector correspond to the respective information to indicate which of the plural attributes in the image are resultatively classified as being attributes of the object.