TasksSotADatasetsPapersMethodsSubmitAbout
Papers With Code 2

A community resource for machine learning research: papers, code, benchmarks, and state-of-the-art results.

Explore

Notable BenchmarksAll SotADatasetsPapersMethods

Community

Submit ResultsAbout

Data sourced from the PWC Archive (CC-BY-SA 4.0). Built by the community, for the community.

Papers/Regressing Robust and Discriminative 3D Morphable Models w...

Regressing Robust and Discriminative 3D Morphable Models with a very Deep Neural Network

Anh Tuan Tran, Tal Hassner, Iacopo Masi, Gerard Medioni

2016-12-15CVPR 2017 7Face RecognitionFace VerificationFace Reconstruction3D Face Reconstruction
PaperPDFCodeCodeCodeCodeCode

Abstract

The 3D shapes of faces are well known to be discriminative. Yet despite this, they are rarely used for face recognition and always under controlled viewing conditions. We claim that this is a symptom of a serious but often overlooked problem with existing methods for single view 3D face reconstruction: when applied "in the wild", their 3D estimates are either unstable and change for different photos of the same subject or they are over-regularized and generic. In response, we describe a robust method for regressing discriminative 3D morphable face models (3DMM). We use a convolutional neural network (CNN) to regress 3DMM shape and texture parameters directly from an input photo. We overcome the shortage of training data required for this purpose by offering a method for generating huge numbers of labeled examples. The 3D estimates produced by our CNN surpass state of the art accuracy on the MICC data set. Coupled with a 3D-3D face matching pipeline, we show the first competitive face recognition results on the LFW, YTF and IJB-A benchmarks using 3D face shapes as representations, rather than the opaque deep feature vectors used by other modern systems.

Results

TaskDatasetMetricValueModel
Facial Recognition and ModellingFlorenceAverage 3D Error1.933DMM-CNN
Facial Recognition and ModellingFlorenceRMSE Cooperative1.97Tran et al.
Facial Recognition and ModellingFlorenceRMSE Indoor2.03Tran et al.
Facial Recognition and ModellingFlorenceRMSE Outdoor1.93Tran et al.
Facial Recognition and ModellingNoW BenchmarkMean Reconstruction Error (mm)2.333DMM-CNN
Facial Recognition and ModellingNoW BenchmarkMedian Reconstruction Error1.843DMM-CNN
Facial Recognition and ModellingNoW BenchmarkStdev Reconstruction Error (mm)2.053DMM-CNN
Face ReconstructionFlorenceAverage 3D Error1.933DMM-CNN
Face ReconstructionFlorenceRMSE Cooperative1.97Tran et al.
Face ReconstructionFlorenceRMSE Indoor2.03Tran et al.
Face ReconstructionFlorenceRMSE Outdoor1.93Tran et al.
Face ReconstructionNoW BenchmarkMean Reconstruction Error (mm)2.333DMM-CNN
Face ReconstructionNoW BenchmarkMedian Reconstruction Error1.843DMM-CNN
Face ReconstructionNoW BenchmarkStdev Reconstruction Error (mm)2.053DMM-CNN
3DFlorenceAverage 3D Error1.933DMM-CNN
3DFlorenceRMSE Cooperative1.97Tran et al.
3DFlorenceRMSE Indoor2.03Tran et al.
3DFlorenceRMSE Outdoor1.93Tran et al.
3DNoW BenchmarkMean Reconstruction Error (mm)2.333DMM-CNN
3DNoW BenchmarkMedian Reconstruction Error1.843DMM-CNN
3DNoW BenchmarkStdev Reconstruction Error (mm)2.053DMM-CNN
3D Face ModellingFlorenceAverage 3D Error1.933DMM-CNN
3D Face ModellingFlorenceRMSE Cooperative1.97Tran et al.
3D Face ModellingFlorenceRMSE Indoor2.03Tran et al.
3D Face ModellingFlorenceRMSE Outdoor1.93Tran et al.
3D Face ModellingNoW BenchmarkMean Reconstruction Error (mm)2.333DMM-CNN
3D Face ModellingNoW BenchmarkMedian Reconstruction Error1.843DMM-CNN
3D Face ModellingNoW BenchmarkStdev Reconstruction Error (mm)2.053DMM-CNN
3D Face ReconstructionFlorenceAverage 3D Error1.933DMM-CNN
3D Face ReconstructionFlorenceRMSE Cooperative1.97Tran et al.
3D Face ReconstructionFlorenceRMSE Indoor2.03Tran et al.
3D Face ReconstructionFlorenceRMSE Outdoor1.93Tran et al.
3D Face ReconstructionNoW BenchmarkMean Reconstruction Error (mm)2.333DMM-CNN
3D Face ReconstructionNoW BenchmarkMedian Reconstruction Error1.843DMM-CNN
3D Face ReconstructionNoW BenchmarkStdev Reconstruction Error (mm)2.053DMM-CNN

Related Papers

ProxyFusion: Face Feature Aggregation Through Sparse Experts2025-09-24DiffClean: Diffusion-based Makeup Removal for Accurate Age Estimation2025-07-17Non-Adaptive Adversarial Face Generation2025-07-16Attributes Shape the Embedding Space of Face Recognition Models2025-07-15Face mask detection project report.2025-07-02On the Burstiness of Faces in Set2025-06-25Identifying Physically Realizable Triggers for Backdoored Face Recognition Networks2025-06-24SELFI: Selective Fusion of Identity for Generalizable Deepfake Detection2025-06-21