Method and system for creating virtual image based deep-learning
A method and system for creating a virtual image based on deep learning according to an embodiment of the present disclosure creates a virtual image based on deep learning by an image application executed by a processor of a computing device, where the method comprises obtaining a plurality of product images with respect to one product; classifying the obtained product images into a plurality of categories according to different pose types; determining a target category from among the plurality of categories for which the virtual image is to be created; creating a virtual image of a first pose type matched to the determined target category based on at least one product image among the plurality of product images; and displaying the created virtual image.
1. A method for creating a virtual image based on deep learning by an image application executed by a processor of a computing device, the method comprising:
obtaining a plurality of product images including one product;
classifying the obtained product images into a plurality of pose type categories according to a pose included in each of the obtained product images;
selecting at least one target pose type category from the plurality of pose type categories to create the virtual image;
creating a virtual image of a pose type corresponding to the selected target pose type category using at least one product image among the plurality of product images by the deep learning; and
outputting the created virtual image, and
wherein the selecting of the at least one target pose type category includes detecting an empty category to which no product image is classified among the plurality of pose type categories, and determining the detected empty category as the target pose type category.
2. The method of claim 1 , wherein the virtual image is created to have the pose type corresponding to the selected target pose type category by inputting the at least one product image and information related to the pose type corresponding to the selected target pose type category to a pre-trained deep learning neural network.
3. The method of claim 1 , wherein the creating of the virtual image of the pose type corresponding to the selected target pose type category includes determining at least one of the plurality of product images as a base image.
4. The method of claim 3 , wherein the determining of the at least one of the plurality of product images as the base image includes determining, as the base image, a product image classified to a pose type category having a highest priority among the plurality of pose type categories, wherein each of the plurality of pose type categories has a respective preconfigured priority.
5. The method of claim 3 , wherein the creating of the virtual image of the pose type corresponding to the selected target pose type category further includes creating a pose semantic label map of the pose type corresponding to the selected target pose type category by inputting the base image and information related to the pose type corresponding to the selected target pose type category to a pre-trained deep learning neural network.
6. The method of claim 5 , wherein the creating of the virtual image of the pose type corresponding to the selected target pose type category further includes creating a base semantic label map which is a semantic label map of the base image.
7. The method of claim 5 , wherein the creating of the virtual image of the pose type corresponding to the selected target pose type category further includes creating the virtual image by inputting the pose semantic label map and the base image to the pre-trained deep learning neural network.
8. The method of claim 5 , wherein the creating of the virtual image of the pose type corresponding to the selected target pose type category includes creating the virtual image having the pose type corresponding to the selected target pose type category by inputting the information related to the pose type corresponding to the selected target pose type category and the pose semantic label map to the pre-trained deep learning neural network to correspond to the base image and the base semantic label map.
9. A system for creating a virtual image based on deep learning, the system comprising:
at least one processor; and
a memory storing instructions for an image application executed by the at least one processor, the instructions for the image application comprising:
obtaining a plurality of product images including one product,
classifying the obtained product images into a plurality of pose type categories according to a pose included in each of the obtained product images,
selecting a target pose type category from the plurality of pose type categories to create the virtual image,
creating a virtual image of a pose type corresponding to the selected target pose type category using at least one product image among the plurality of product images by the deep learning, and
outputting the created virtual image, and
wherein the instructions for the image application include detecting an empty category to which no product image is classified among the plurality of pose type categories, and determining the detected empty category as the target pose type category.
10. The system of claim 9 , wherein the instructions for the image application include determining at least one of the plurality of product images as a base image.
11. The system of claim 10 , wherein the instructions for the image application include determining, as the base image, a product image classified to a pose type category having a highest priority among the plurality of pose type categories, wherein each of the plurality of pose type categories has a respective preconfigured priority.
12. The system of claim 11 , wherein the instructions for the image application include creating a pose semantic label map of the pose type corresponding to the selected target pose type category by inputting the base image and information related to the pose type corresponding to the selected target pose type category to a pre-trained deep learning neural network.
13. The system of claim 12 , wherein the instruction for the image application include creating a base semantic label map which is a semantic label map of the base image.
14. The system of claim 13 , wherein the instructions for the image application include creating the virtual image by inputting the pose semantic label map and the base image to the pre-trained deep learning neural network.
15. The system of claim 14 , wherein the instructions for the image application include creating the virtual image having the pose type corresponding to the selected target pose type category by inputting the information related to the pose type corresponding to the selected target pose type category and the pose semantic label map to the pre-trained deep learning neural network to correspond to the base image and the base semantic label map.