Creation of Digital Virtual Human Based on GAN2Shape and Kinect 2.0

This study aimed at the problem of the low recognition rate of facial expressions cause by complex environments. Hence, the study examined, digital virtual humans through GAN2Shape and Kinect 2.0 to obtain high-quality three-dimensional (3D) images of human faces. First, we captured and recognized facial expressions through Kinect. Second, we generated high-quality 3D images of the facial expressions through GAN2Shape and transferred the images to Unity 3D for front and back-end development to generate a facial model. Third, we recognized and captured motions through Kinect. Finally, we combined the facial model with the motions to form the virtual human model. Experimental results show that the virtual human model generated through this method has a clear face and a good effect in motion capture.