Portrait Segmentation by Deep Refinement of Image Matting

Portrait segmentation is becoming a hot topic nowadays. In this paper we propose a novel framework to cope with the high precision requirements that portrait segmentation demands on boundary area by deep refinement of the portrait matting. Our approach introduces three novel techniques. First, a trimap is proposed by fusing information coming from two well-known techniques for image segmentation, i.e., Mask R-CNN and DensePose. Second, an alpha matting algorithm runs over the previous trimap generate. From this mate result we generate a couple of masks, one of them boundary-sensitive kernel, called boundary and the other one inside-sensitive kernel called leftover. Third, we refine the portrait by a pre-trained CNN-based model, followed by a transposed convolution. We have evaluated our approach on the PFCN dataset as well as the portrait images collected from COCO dataset. Experimental results demonstrate the better performance of our algorithm over previous methods.