A 3D Face Model for Pose and Illumination Invariant Face Recognition

By Pascal Paysan, Reinhard Knothe, Brian Amberg, Sami Romdhani, Thomas Vetter. 2009

1,562 citations recorded by the source corpus.

Subfields: Computer Vision and Pattern Recognition

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Vision and Pattern Recognition

Citations in the corpus

InfoGAN: Interpretable Representation Learning by Information Maximizing Generative Adversarial Nets (2016, methodology): “Next we evaluate InfoGAN on two datasets of 3D images: faces [28] and chairs [29], on which DC-IGN was shown to learn highly interpretable graphics codes.”
Deep Convolutional Inverse Graphics Network (2015, methodology): “In all our experiments, we train using the synthetic data obtained from the 3DMM model and test it out on real face data with ground truth 3D annotations (geometry and texture) obtained from [23].”
Learning to Discover Cross-Domain Relations with Generative Adversarial Networks (2017, methodology): “All computations were conducted on a single machine with an Nvidia Titan X Pascal GPU and an Intel(R) Xeon(R) E5-1620 CPU.”
Visual interpretability for deep learning: a survey (2018, methodology): “, 2011], the 3D face dataset [Paysan et al., 2009], and the 3D chair dataset [Aubry et al.”
Disentangling by Factorising (2018, background): “2) 3D Shapes data: 480,000 RGB 64× 64× 3 images of 3D shapes with ground truth factors: shape[4], scale[8], orientation[15], floor colour[10], wall colour[10], object colour[10] ii) unknown generative factors: 3) 3D Faces (Paysan et al., 2009): 239,840 grey-scale 64× 64 images of 3D Faces.”
Isolating Sources of Disentanglement in VAEs (2018, methodology): “We perform quantitative evaluations with two datasets, a dataset of 2D shapes [39] and a dataset of synthetic 3D faces [40].”
Generating 3D faces using Convolutional Mesh Autoencoders (2018, methodology): “To date, the Basel Face Model (BFM) [36], i.”
i-RevNet: Deep Invertible Networks (2018, methodology): “Results are displayed in the Figure 5. dataset et We selected images from the basel face (Paysan al., 2009), describable texture dataset (Cimpoi et al., 2014) and imagenet.”
Understanding Visual Concepts with Continuation Learning (2016, methodology): “We trained the model on faces generated from the Basel face model (Paysan et al. 2009) and prepared as in (Kulkarni et al. 2015).”
Recent Advances in Autoencoder-Based Representation Learning (2018, background): “…color and facial attributes such as glasses), and synthetic data sets of geometric 2D shapes or rendered 3D shapes (e.g., 2D Shapes [2], 3D Shapes [3], 3D Faces [39], 3D Chairs [40]) for which the data generative process and hence the ground truth factors are known (see Figure 4 for an example).”
Picture: A probabilistic programming language for scene perception (2015, methodology): “(Even better results can be obtained using the deep convolutional inverse graphics network [25] instead of the CNN.) We evaluated the program on a held-out test set of 2D projected images of 3D laser scanned data (dataset from [36]).”
Synthesizing Programs for Images using Reinforced Adversarial Learning (2018, background): “The idea of inverting simulators to interpret images has been explored extensively in recent years (Nair et al., 2008; Paysan et al., 2009; Mansinghka et al., 2013; Loper & Black, 2014; Kulkarni et al., 2015a; Jampani et al., 2015).”
FaceID-GAN: Learning a Symmetry Three-Player GAN for Identity-Preserving Face Synthesis (2018, background): “where S is a mean shape of a 3D face, Aid and Aexp are the PCA bases for shape [23] and expression [4] respectively.”
Bayesian CP Factorization of Incomplete Tensors with Automatic Rank Determination (2014, methodology): “For illustration, we use the dataset of 3D Basel Face Model [37] that contains an ensemble of facial images of 10 people, each rendered in 9 different poses under 3 different illuminations.”
Disentangled and Controllable Face Image Generation via 3D Imitative-Contrastive Learning (2020, methodology): “To learn how a face image should be generated following the desired properties, we incorporate a 3DMM model [33] and train the generator to imitate the rendered 3D faces.”
End-to-End 3D Face Reconstruction with Deep Neural Networks (2017, methodology): “We use two 3D facial shape models, namely the BFM model proposed by [20] and the AFM model proposed by [12].”
Disentangled Representations in Neural Models (2016, methodology): “Various statistical shape models trained on 3D scanner data such as faces have the same multivariate normal latent distribution (Paysan et al. 2009).”
Extreme 3D Face Reconstruction: Seeing Through Occlusions (2017, methodology): “BFM, used to represent the foundation shape, provides a complete face representation and their vertices allow for easy indexing of symmetric face regions.”
Computer Vision and Image Understanding (2016, methodology): “To this end, we generated a dataset of synthetic 3D faces simulating weight changes using a parametric deformable model, namely the Basel Face Model ( Paysan et al., 2009 ).”