2017/08/23 by Anil Bas, Patrik Huber, William A. P. Smith +2 · 70 citations
Computer Science · Engineering · #3D Shape Modeling and Analysis #3d model #Artificial intelligence #Computer science #Computer vision #Convolutional neural network #Engineering #Face recognition and analysis #Generative Adversarial Networks and Image Synthesis #Pattern recognition (psychology) #Solid modeling #Transformer #Voltage #acm:68T45 #cs.CV #cs.LG #msc:68T45
paper · pdf · doi:10.1109/iccvw.2017.110
Accepted to ICCV 2017 2nd Workshop on Geometry Meets Deep Learning
arxiv created 2017/08/23 · openalex publication_date 2017/10/01 · arxiv updated 2018/04/20 · openalex created_date 2025/10/10 · openalex updated_date 2026/08/05
In this paper, we show how a 3D Morphable Model (i.e. a statistical model of the 3D shape of a class of objects such as faces) can be used to spatially transform input data as a module (a 3DMM-STN) within a convolutional neural network. This is an extension of the original spatial transformer network in that we are able to interpret and normalise 3D pose changes and self-occlusions. The trained localisation part of the network is independently useful since it learns to fit a 3D morphable model to a single image. We show that the localiser can be trained using only simple geometric loss functions on a relatively small dataset yet is able to perform robust normalisation on highly uncontrolled images including occlusion, self-occlusion and large pose changes.