• Corpus ID: 228064212

MorphGAN: One-Shot Face Synthesis GAN for Detecting Recognition Bias

@inproceedings{Ruiz2020MorphGANOF,
  title={MorphGAN: One-Shot Face Synthesis GAN for Detecting Recognition Bias},
  author={Nataniel Ruiz and Barry-John Theobald and Anurag Ranjan and Ahmed Abdelaziz and Nicholas Apostoloff},
  booktitle={British Machine Vision Conference},
  year={2020}
}
To detect bias in face recognition networks, it can be useful to probe a network under test using samples in which only specific attributes vary in some controlled way. However, capturing a sufficiently large dataset with specific control over the attributes of interest is difficult. In this work, we describe a simulator that applies specific head pose and facial expression adjustments to images of previously unseen people. The simulator first fits a 3D morphable model to a provided image… 
[PDF] Semantic Reader
6 Citations
Background Citations
4

Figures and Tables from this paper

6 Citations

Controllable and Guided Face Synthesis for Unconstrained Face Recognition

A controllable face synthesis model that can mimic the distribution of target datasets in a style latent space with precise control over the diversity and degree of synthesis and yields significant performance gains on unconstrained bench-marks.

Simulated Adversarial Testing of Face Recognition Models

This work proposes a framework for learning how to test machine learning algorithms using simulators in an adversarial manner in order to find weaknesses in the model before deploying it in critical scenarios, and presents a method to find adversarial regions as opposed to the typical adversarial points found in the adversarial example literature.

Supplementary Material: Simulated Adversarial Testing of Face Recognition Models

  • Computer Science
  • 2022
In order to explore samples generated by simulated adversarial testing and other simulated testing techniques, the shape and texture components of the authors' samples areprojected onto a plane of two components to give a strong intuition over what features affect network performance.

Finding Differences Between Transformers and ConvNets Using Counterfactual Simulation Testing

Counterfactual Simulation Testing is presented, a counterfactual framework that allows for a fair comparison of the robustness of recently released, state-of-the-art Convolutional Neural Networks and Vision Transformers, with respect to naturalistic variations of object pose, scale, viewpoint, lighting and occlusions.

Human Body Measurement Estimation with Adversarial Augmentation

A Body Measurement network (BMnet) for estimating 3D anthropomorphic measurements of the human body shape from silhouette images is presented, augmented with a novel adversarial body simulator (ABS) that finds and synthesizes challenging body shapes.

3D human tongue reconstruction from single “in-the-wild” images

This work presents the first, to the best of their knowledge, end-to-end trainable pipeline that accurately reconstructs the 3D face together with the tongue, and makes this pipeline robust in “in-the-wild” images by introducing a novel GAN method tailored for 3D tongue surface generation.

References

SHOWING 1-10 OF 64 REFERENCES

Cross-Domain Face Synthesis using a Controllable GAN

A cross-domain face synthesis approach is proposed that integrates a new Controllable GAN (C-GAN) that employs an off-the-shelf 3D face model as a simulator to generate facial images under various poses and relies on an additional adversarial game as a third player to preserve the identity and specific facial attributes of the refined images.

UV-GAN: Adversarial Facial UV Map Completion for Pose-Invariant Face Recognition

A framework for training Deep Convolutional Neural Network (DCNN) to complete the facial UV map extracted from in-the-wild images, and devise a meticulously designed architecture that combines local and global adversarial DCNNs to learn an identity-preserving facial UV completion model.

GANimation: Anatomically-aware Facial Animation from a Single Image

A novel GAN conditioning scheme based on Action Units (AU) annotations, which describes in a continuous manifold the anatomical facial movements defining a human expression, and proposes a fully unsupervised strategy to train the model, that only requires images annotated with their activated AUs.

Learning to Regress 3D Face Shape and Expression From an Image Without 3D Supervision

To train a network without any 2D-to-3D supervision, RingNet is presented, which learns to compute 3D face shape from a single image and achieves invariance to expression by representing the face using the FLAME model.

Analyzing and Reducing the Damage of Dataset Bias to Face Recognition With Synthetic Data

This study demonstrates the large potential of synthetic data for analyzing and reducing the negative effects of dataset bias on deep face recognition systems and shows that current neural network architectures cannot disentangle face pose and facial identity, which limits their generalization ability.

Face Generation for Low-Shot Learning Using Generative Adversarial Networks

A generator from the Generative Adversarial Network is adapted to increase the size of training dataset to improve the accuracy and robustness of face recognition and it is concluded that the proposed algorithm for generating faces is effective in improving the identification accuracy and coverage.

Photo-Realistic Facial Details Synthesis From Single Image

We present a single-image 3D face synthesis technique that can handle challenging facial expressions while recovering fine geometric details. Our technique employs expression analysis for proxy face…

GIF: Generative Interpretable Faces

This work conditiones a generative 2D face model named GIF (Generative Interpretable Faces) that offers FLAME’s parametric control and performs an AMT based perceptual study to quantitatively and qualitatively evaluate how well GIF follows its conditioning.

3D Guided Fine-Grained Face Manipulation

This work presents a method for fine-grained face manipulation that can synthesize another arbitrary expression by the same person by first fitting a 3D face model and then disentangling the face into a texture and a shape.

Generating 3D faces using Convolutional Mesh Autoencoders

This work introduces a versatile model that learns a non-linear representation of a face using spectral convolutions on a mesh surface and shows that, replacing the expression space of an existing state-of-the-art face model with this model, achieves a lower reconstruction error.
...