Abstract: Provided is a system and method for creating and providing a mosaic image based on an image tag-word by a mosaic service providing server, which includes: determining each tag-word for each image and classifying a plurality of images according to the determined tag-word; determining a target image among the plurality of images; providing a pixel image selection interface for selecting a pixel image for mosaicizing the determined target image based on the tag-words of the plurality of images; and creating a mosaic image for the target image based on the pixel image selected through the pixel image selection interface.

Abstract: A deep learning-based virtual hair dyeing method is performed by an image editing server including a deep learning neural network and generative adversarial networks (GANs). The method comprises obtaining an original image and a desired hair color; generating a semantic map by providing the original image to the deep learning neural network; generating a trimap automatically with respect to the original image based on the semantic map; generating an alpha mask by performing alpha matting on the original image based on the trimap; obtaining a transformed image by applying the original image to the GANs; and obtaining a virtual dyeing image by synthesizing the original image and the obtained transformed image through the alpha mask.

Abstract: Provided is a system and method for creating and providing a mosaic image based on an image tag-word by a mosaic service providing server, which includes: determining each tag-word for each image and classifying a plurality of images according to the determined tag-word; determining a target image among the plurality of images; providing a pixel image selection interface for selecting a pixel image for mosaicizing the determined target image based on the tag-words of the plurality of images; and creating a mosaic image for the target image based on the pixel image selected through the pixel image selection interface.

Abstract: A deep learning-based virtual hair dyeing method is performed by an image editing server including a deep learning neural network and generative adversarial networks (GANs). The method comprises obtaining an original image and a desired hair color; generating a semantic map by providing the original image to the deep learning neural network; generating a trimap automatically with respect to the original image based on the semantic map; generating an alpha mask by performing alpha matting on the original image based on the trimap; obtaining a transformed image by applying the original image to the GANs; and obtaining a virtual dyeing image by synthesizing the original image and the obtained transformed image through the alpha mask.