This repo contains the code for generating the orientation images in the paper HairNet: Single-View Hair Reconstruction using Convolutional Neural Networks.
I want to know that whether the body image is derived from the segmentation performed on the rgb image or by projecting the half body model into image after aligned. If it is the latter, how to align the half body model to the rgb image