Deep Models and Shortwave Infrared Information to Detect Face Presentation Attacks

This paper addresses the problem of face presentation attack detection using different image modalities. In particular, the usage of short wave infrared (SWIR) imaging is considered. Face presentation attack detection is performed using recent models based on Convolutional Neural Networks using only carefully selected SWIR image differences as input. Conducted experiments show superior performance over similar models acting on either color images or on a combination of different modalities (visible, NIR, thermal and depth), as well as on a SVM-based classifier acting on SWIR image differences. Experiments have been carried on a new public and freely available database, containing a wide variety of attacks. Video sequences have been recorded thanks to several sensors resulting in 14 different streams in the visible, NIR, SWIR and thermal spectra, as well as depth data. The best proposed approach is able to almost perfectly detect all impersonation attacks while ensuring low \bona classification errors. On the other hand, obtained results show that obfuscation attacks are more difficult to detect. We hope that the proposed database will foster research on this challenging problem. Finally, all the code and instructions to reproduce presented experiments is made available to the research community.

A Cross-database Study of Voice Presentation Attack Detection

Sébastien Marcel

Despite an increasing interest in speaker recognition technologies, a significant obstacle still hinders their wide deployment --- their high vulnerability to spoofing or presentation attacks. These attacks can be easy to perform. For instance, if an attacker has access to a speech sample from a target user, he/she can replay it using a loudspeaker or a smartphone to the recognition system during the authentication process. The ease of executing presentation attacks and the fact that no technical knowledge of the biometric system is required makes these attacks especially threatening in practical application. Therefore, late research focuses on collecting data databases with such attacks and on development of presentation attack detection (PAD) systems. In this chapter, we present an overview of the latest databases and the techniques to detect presentation attacks. We consider several prominent databases that contain bona fide and attack data, including: ASVspoof 2015, ASVspoof 2017, AVspoof, voicePA, and BioCPqD-PA (the only proprietary database). Using these databases, we focus on the performance of PAD systems in the cross-database scenario or in the presence of 'unknown' (not available during training) attacks, as these scenarios are closer to practice, when pre-trained systems need to detect attacks in unforeseen conditions. We first present and discuss the performance of PAD systems based on handcrafted features and traditional Gaussian mixture model (GMM) classifiers. We then demonstrate whether the score fusion techniques can improve the performance of PADs. We also present some of the latest results of using neural networks for presentation attack detection. The experiments show that PAD systems struggle to generalize across databases and mostly unable to detect unknown attacks, with systems based on neural networks demonstrating better performance compared to the systems based on handcraft features.

Springer2018

Investigating Depth Domain Adaptation for Efficient Human Pose Estimation

Olivier Canévet, Jean-Marc Odobez

Convolutional Neural Networks (CNN) are the leading models for human body landmark detection from RGB vision data. However, as such models require high computational load, an alternative is to rely on depth images which, due to their more simple nature, can allow the use of less complex CNNs and hence can lead to a faster detector. As learning CNNs from scratch requires large amounts of labeled data, which are not always available or expensive to obtain, we propose to rely on simulations and synthetic examples to build a large training dataset with precise labels. Nevertheless, the final performance on real data will suffer from the mismatch between the training and test data, also called domain shift between the source and target distributions. Thus in this paper, our main contribution is to investigate the use of unsupervised domain adaptation techniques to fill the gap in performance introduced by these distribution differences. The challenge lies in the important noise differences (not only gaussian noise, but many missing values around body limbs) between synthetic and real data, as well as the fact that we address a regression task rather than a classification one. In addition, we introduce a new public dataset of synthetically generated depth images to cover the cases of multi-person pose estimation. Our experiments show that domain adaptation provides some improvement, but that further network fine-tuning with real annotated data is worth including to supervise the adaptation process.

2018

Evaluating and Interpreting Deep Convolutional Neural Networks via Non-negative Matrix Factorization

Edo Collins

With ever greater computational resources and more accessible software, deep neural networks have become ubiquitous across industry and academia. Their remarkable ability to generalize to new samples defies the conventional view, which holds that complex, over-parameterized networks would be prone to overfitting. This apparent discrepancy is exacerbated by our inability to inspect and interpret the high-dimensional, non-linear, latent representations they learn, which has led many to refer to neural networks as black-boxes''. The Law of Parsimony states that simpler solutions are more likely to be correct than complex ones''. Since they perform quite well in practice, a natural question to ask, then, is in what way are neural networks simple? We propose that compression is the answer. Since good generalization requires invariance to irrelevant variations in the input, it is necessary for a network to discard this irrelevant information. As a result, semantically similar samples are mapped to similar representations in neural network deep feature space, where they form simple, low-dimensional structures. Conversely, a network that overfits relies on memorizing individual samples. Such a network cannot discard information as easily. In this thesis we characterize the difference between such networks using the non-negative rank of activation matrices. Relying on the non-negativity of rectified-linear units, the non-negative rank is the smallest number that admits an exact non-negative matrix factorization. We derive an upper bound on the amount of memorization in terms of the non-negative rank, and show it is a natural complexity measure for rectified-linear units. With a focus on deep convolutional neural networks trained to perform object recognition, we show that the two non-negative factors derived from deep network layers decompose the information held therein in an interpretable way. The first of these factors provides heatmaps which highlight similarly encoded regions within an input image or image set. We find that these networks learn to detect semantic parts and form a hierarchy, such that parts are further broken down into sub-parts. We quantitatively evaluate the semantic quality of these heatmaps by using them to perform semantic co-segmentation and co-localization. In spite of the convolutional network we use being trained solely with image-level labels, we achieve results comparable or better than domain-specific state-of-the-art methods for these tasks. The second non-negative factor provides a bag-of-concepts representation for an image or image set. We use this representation to derive global image descriptors for images in a large collection. With these descriptors in hand, we perform two variations content-based image retrieval, i.e. reverse image search. Using information from one of the non-negative matrix factors we obtain descriptors which are suitable for finding semantically related images, i.e., belonging to the same semantic category as the query image. Combining information from both non-negative factors, however, yields descriptors that are suitable for finding other images of the specific instance depicted in the query image, where we again achieve state-of-the-art performance.