Face Semantic Segmentation and Enhanced Attribute Description on Portrait Robot

Traditionally, image drawing robot draws a portrait only by generated image, without concerning the certain order of people's facial features. In this paper, a method of attribute enhanced description based on semantic segmentation of face image is proposed. The semantics of face image is segmented by BiSeNet, and the data sets of different feature organs are constructed. We use the enhanced attribute module to generate the description of the relevant attributes for each feature organ image. Unlike the previous soft biological features of human face, this method can more intelligently realize the description of feature attributes. The experimental results show that our method can obtain better organ localization and enhanced attribute description. Combining our existing drawing robot platform with our drawing strategy based on enhanced attribute description, it has a good promotion effect on humanoid drawing of robots.