Andreas Schneider

Adam Kortylewski, Bernhard Egger, Andreas Morel-Forster et al.

It is well known that deep learning approaches to face recognition and facial landmark detection suffer from biases in modern training datasets. In this work, we propose to use synthetic face images to reduce the negative effects of dataset biases on these tasks. Using a 3D morphable face model, we generate large amounts of synthetic face images with full control over facial shape and color, pose, illumination, and background. With a series of experiments, we extensively test the effects of priming deep nets by pre-training them with synthetic faces. We observe the following positive effects for face recognition and facial landmark detection tasks: 1) Priming with synthetic face images improves the performance consistently across all benchmarks because it reduces the negative effects of biases in the training data. 2) Traditional approaches for reducing the damage of dataset bias, such as data augmentation and transfer learning, are less effective than training with synthetic faces. 3) Using synthetic data, we can reduce the size of real-world datasets by 75% for face recognition and by 50% for facial landmark detection while maintaining performance. Thus, offering a means to focus the data collection process on less but higher quality data.

Adam Kortylewski, Andreas Schneider, Thomas Gerig et al.

Recent advances in deep learning have significantly increased the performance of face recognition systems. The performance and reliability of these models depend heavily on the amount and quality of the training data. However, the collection of annotated large datasets does not scale well and the control over the quality of the data decreases with the size of the dataset. In this work, we explore how synthetically generated data can be used to decrease the number of real-world images needed for training deep face recognition systems. In particular, we make use of a 3D morphable face model for the generation of images with arbitrary amounts of facial identities and with full control over image variations, such as pose, illumination, and background. In our experiments with an off-the-shelf face recognition software we observe the following phenomena: 1) The amount of real training data needed to train competitive deep face recognition systems can be reduced significantly. 2) Combining large-scale real-world data with synthetic data leads to an increased performance. 3) Models trained only on synthetic data with strong variations in pose, illumination, and background perform very well across different datasets even without dataset adaptation. 4) The real-to-virtual performance gap can be closed when using synthetic data for pre-training, followed by fine-tuning with real-world images. 5) There are no observable negative effects of pre-training with synthetic data. Thus, any face recognition system in our experiments benefits from using synthetic face images. The synthetic data generator, as well as all experiments, are publicly available.

Adam Kortylewski, Bernhard Egger, Andreas Schneider et al.

It is unknown what kind of biases modern in the wild face datasets have because of their lack of annotation. A direct consequence of this is that total recognition rates alone only provide limited insight about the generalization ability of a Deep Convolutional Neural Networks (DCNNs). We propose to empirically study the effect of different types of dataset biases on the generalization ability of DCNNs. Using synthetically generated face images, we study the face recognition rate as a function of interpretable parameters such as face pose and light. The proposed method allows valuable details about the generalization performance of different DCNN architectures to be observed and compared. In our experiments, we find that: 1) Indeed, dataset bias has a significant influence on the generalization performance of DCNNs. 2) DCNNs can generalize surprisingly well to unseen illumination conditions and large sampling gaps in the pose variation. 3) Using the presented methodology we reveal that the VGG-16 architecture outperforms the AlexNet architecture at face recognition tasks because it can much better generalize to unseen face poses, although it has significantly more parameters. 4) We uncover a main limitation of current DCNN architectures, which is the difficulty to generalize when different identities to not share the same pose variation. 5) We demonstrate that our findings on synthetic data also apply when learning from real-world data. Our face image generator is publicly available to enable the community to benchmark other DCNN architectures.