Section: New Results

Robust Multilinear Model Learning Framework for 3D Faces

Statistical models are widely used to represent the variations of 3D human faces. Multilinear models in particular are common as they decouple shape changes due to identity and expression. Existing methods to learn a multilinear face model degrade if not every person is captured in every expression, if face scans are noisy or partially occluded, if expressions are erroneously labeled, or if the vertex correspondence is inaccurate. These limitations impose requirements on the training data that disqualify large amounts of available 3D face data from being usable to learn a multilinear model. To overcome this, we have developed an effective framework to robustly learn a multilinear model from 3D face databases with missing data, corrupt data, wrong semantic correspondence, and inaccurate vertex correspondence. To achieve this robustness to erroneous training data, our framework jointly learns a multilinear model and fixes the data. This framework is significantly more efficient than prior methods based on linear statistical models. This work was presented at CVPR 2016 [8].

Figure 5. Overview of our robust multilinear model (RMM) learning framework that is robust to missing data (purple), corrupt data (brown), wrong semantic correspondence (green), and inaccurate vertex correspondence (gray).