| CPC G06T 11/001 (2013.01) [G06T 11/60 (2013.01); G06V 10/771 (2022.01); G06V 10/82 (2022.01); G06T 2210/44 (2013.01)] | 16 Claims |
|
1. A caricature generating method, comprising:
providing a generation network comprising a plurality of layers connected in series including coarse layers of lowest resolutions and pre-trained to be suitable for synthesizing a shape of a caricature and fine layers of highest resolutions and pre-trained to be suitable for tuning a texture of the caricature;
applying input feature maps representing an input facial photograph to the coarse layers to generate shape feature maps and deforming the shape feature maps by shape exaggeration blocks each comprising at least one convolutional layer to generate deformed shape feature maps;
applying the deformed shape feature maps to the fine layers to change a texture represented by the deformed shape feature maps and generate output feature maps; and
generating a caricature image according to the output feature map,
wherein the providing the generation network comprises:
preparing a plurality of series-connected caricature generation learning layers, training the plurality of series-connected caricature generation learning layers with caricature images, preparing a plurality of series-connected photo reconstruction learning layers, and training the plurality of series-connected photo reconstruction learning layers with photographs; and
adopting caricature generation learning layers of the lowest resolutions as the coarse layers, adopting photo reconstruction learning layers of the highest resolutions as the fine layers, and combining the coarse layers and the fine layers to configure the generation network.
|