FAUST: Dataset and Evaluation for 3D Mesh Registration

By Federica Bogo, Javier Romero, Matthew Maverick Loper, Michael J. Black. 2014

831 citations recorded by the source corpus.

Subfields: Computational Mechanics, Computer Vision and Pattern Recognition, Evaluation & Benchmarks

How researchers see this paper

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

In Computational Mechanics

In Computer Vision and Pattern Recognition

In Evaluation & Benchmarks

Citations in the corpus

Geometric Deep Learning on Graphs and Manifolds Using Mixture Model CNNs (2016, methodology): “We reproduced verbatim the experiments of [32, 7] on the FAUST humans dataset [5], comparing to the methods reported therein.”
Fast Graph Representation Learning with PyTorch Geometric (2019, methodology): “In addition, we provide embedded datasets like MNIST superpixels (Monti et al., 2017), FAUST (Bogo et al., 2014), ModelNet10/40 (Wu et al., 2015), ShapeNet (Chang et al., 2015), COMA (Ranjan et al., 2018), and the PCPNet dataset from Guerrero et al. (2018).”
SplineCNN: Fast Geometric Deep Learning with Continuous B-Spline Kernels (2017, methodology): “We use the FAUST dataset [1], containing 10 scanned human shapes in 10 different poses, resulting in a total of 100 non-watertight meshes with 6,890 nodes each.”
Pix3D: Dataset and Methods for Single-Image 3D Shape Modeling (2018, background): “For decades, researchers have been building datasets of 3D objects, either as a repository of 3D CAD models [4, 5, 49] or as images of 3D shapes with pose annotations [34, 47].”
A Papier-Mache Approach to Learning 3D Surface Generation (2018, positive): “re for our method. 5.4. Additional applications Deformable shapes. We ran an experiment on human shape to show that our method is also suitable for reconstructing deformable shapes. The FAUST dataset [3] is a collection of meshes representing several humans in different poses. We used 250 shapes for”
Octree Generating Networks: Efficient Convolutional Architectures for High-resolution 3D Outputs (2017, methodology): “et-cars A subset of ShapeNet-all consisting of 7497 car models. FAUST A collection of 300 high-resolution scans of human bodies of 10 different people in 30 different poses from the MPI-FAUST dataset [1]. BlendSwap A dataset of 4 scenes we manually collected from blendswap.com. All datasets were vox”
Learning shape correspondence with anisotropic convolutional neural networks (2016, methodology): “Full mesh correspondence We used the FAUST humans dataset [3], containing 100 meshes of 10 scanned subjects, each in 10 different poses.”
Dense Human Body Correspondences Using Convolutional Networks (2015, methodology): “Dense correspondences between shapes with larger deformations can be obtained reliably using statistical models of human shapes [4, 7], but the subject has to be naked [6].”
Deep Functional Maps: Structured Prediction for Dense Shape Correspondence (2017, methodology): “We carried out experiments on real 3D acquisitions using the recent FAUST benchmark [7].”
Deformable Shape Completion with Graph Convolutional Autoencoders (2017, methodology): “We train the model directly on the 3 × N input meshes from the DFAUST dataset as described above; the sparse adjacency matrices (we use a vertex 2-ring as the neighborhood size) are passed as side information to deﬁne the graph convolutional layers.”
Monocular human pose estimation: A survey of deep learning-based methods (2020, methodology): “3D human body shape scans, such as SCAPE (Anguelov et al., 2005), INRIA4D (INRIA4D, accessed on 2019) and FAUST (Bogo et al., 2014, 2017), 3D human body surface cloud points with time of ﬂight (TOF) depth sensors (Shahroudy et al., 2016), 3D human body reﬂective markers capture with motion capture s”
FeaStNet: Feature-Steered Graph Convolutions for 3D Shape Analysis (2017, methodology): “We follow the experimental setup in [2, 14, 15] based on the FAUST human shape dataset [1].”
Parallel Transport Convolution: A New Tool for Convolutional Neural Networks on Manifolds (2018, methodology): “We evaluated this model the Faust dataset which contain 100 real world scans (each with n = 6890 points) of 10 individuals in 10 poses [2].”
Shape‐from‐Operator: Recovering Shapes from Intrinsic Operators (2015, methodology): “For our experiments, we used shapes from standard shape collections including AIM@SHAPE, TOSCA [BBK08], and FAUST [BRLB14] resampled to 2K–5K vertices as well as additional synthetic shapes.”
3D-CODED: 3D Correspondences by Deep Deformation (2018, methodology): “Furthermore, while our network can take advantage of known correspondences between the template and the example shapes, which are typically available when they have been generated using some parametric model [6,45], we show it can also be trained without correspondence supervision.”
Dreaming More Data: Class-dependent Distributions over Diffeomorphisms for Learned Data Augmentation (2015, background): “• Mesh alignment is also standard pre-processing step in the analysis of three-dimensional meshes [1, 3, 9, 25].”
DiffusionNet: Discretization Agnostic Learning on Surfaces (2020, methodology): “We study discretization invariance using a popular formulation of the shape correspondence task on the FAUST human dataset [Bogo et al. 2014], where each vertex of a mesh is to be labelled with the corresponding vertex on a template mesh.”
Learning a Deep Model for Human Action Recognition from Novel Viewpoints (2016, methodology): “[57] developed the FAUST dataset containing full 3D human body”