Showing 1–10 of 10 results for author: Geissbühler, D

Synthetic Face Datasets Generation via Latent Space Exploration from Brownian Identity Diffusion

Authors: David Geissbühler, Hatef Otroshi Shahreza, Sébastien Marcel

Abstract: Face Recognition (FR) models are trained on large-scale datasets, which have privacy and ethical concerns. Lately, the use of synthetic data to complement or replace genuine data for the training of FR models has been proposed. While promising results have been obtained, it still remains unclear if generative models can yield diverse enough data for such tasks. In this work, we introduce a new met… ▽ More Face Recognition (FR) models are trained on large-scale datasets, which have privacy and ethical concerns. Lately, the use of synthetic data to complement or replace genuine data for the training of FR models has been proposed. While promising results have been obtained, it still remains unclear if generative models can yield diverse enough data for such tasks. In this work, we introduce a new method, inspired by the physical motion of soft particles subjected to stochastic Brownian forces, allowing us to sample identities distributions in a latent space under various constraints. With this in hands, we generate several face datasets and benchmark them by training FR models, showing that data generated with our method exceeds the performance of previously GAN-based datasets and achieves competitive performance with state-of-the-art diffusion-based synthetic datasets. We also show that this method can be used to mitigate leakage from the generator's training set and explore the ability of generative models to generate data beyond it. △ Less

Submitted 30 April, 2024; originally announced May 2024.

Comments: 17 pages, 7 figures, 10 tables

\textit{sweet} -- An Open Source Modular Platform for Contactless Hand Vascular Biometric Experiments

Authors: David Geissbühler, Sushil Bhattacharjee, Ketan Kotwal, Guillaume Clivaz, Sébastien Marcel

Abstract: Current finger-vein or palm-vein recognition systems usually require direct contact of the subject with the apparatus. This can be problematic in environments where hygiene is of primary importance. In this work we present a contactless vascular biometrics sensor platform named \sweet which can be used for hand vascular biometrics studies (wrist-, palm- and finger-vein) and surface features such a… ▽ More Current finger-vein or palm-vein recognition systems usually require direct contact of the subject with the apparatus. This can be problematic in environments where hygiene is of primary importance. In this work we present a contactless vascular biometrics sensor platform named \sweet which can be used for hand vascular biometrics studies (wrist-, palm- and finger-vein) and surface features such as palmprint. It supports several acquisition modalities such as multi-spectral Near-Infrared (NIR), RGB-color, Stereo Vision (SV) and Photometric Stereo (PS). Using this platform we collect a dataset consisting of the fingers, palm and wrist vascular data of 120 subjects and develop a powerful 3D pipeline for the pre-processing of this data. We then present biometric experimental results, focusing on Finger-Vein Recognition (FVR). Finally, we discuss fusion of multiple modalities, such palm-vein combined with palm-print biometrics. The acquisition software, parts of the hardware design, the new FV dataset, as well as source-code for our experiments are publicly available for research purposes. △ Less

Submitted 14 April, 2024; originally announced April 2024.

MSC Class: I.2; I.4; I.5

A Comprehensive Evaluation on Multi-channel Biometric Face Presentation Attack Detection

Authors: Anjith George, David Geissbuhler, Sebastien Marcel

Abstract: The vulnerability against presentation attacks is a crucial problem undermining the wide-deployment of face recognition systems. Though presentation attack detection (PAD) systems try to address this problem, the lack of generalization and robustness continues to be a major concern. Several works have shown that using multi-channel PAD systems could alleviate this vulnerability and result in more… ▽ More The vulnerability against presentation attacks is a crucial problem undermining the wide-deployment of face recognition systems. Though presentation attack detection (PAD) systems try to address this problem, the lack of generalization and robustness continues to be a major concern. Several works have shown that using multi-channel PAD systems could alleviate this vulnerability and result in more robust systems. However, there is a wide selection of channels available for a PAD system such as RGB, Near Infrared, Shortwave Infrared, Depth, and Thermal sensors. Having a lot of sensors increases the cost of the system, and therefore an understanding of the performance of different sensors against a wide variety of attacks is necessary while selecting the modalities. In this work, we perform a comprehensive study to understand the effectiveness of various imaging modalities for PAD. The studies are performed on a multi-channel PAD dataset, collected with 14 different sensing modalities considering a wide range of 2D, 3D, and partial attacks. We used the multi-channel convolutional network-based architecture, which uses pixel-wise binary supervision. The model has been evaluated with different combinations of channels, and different image qualities on a variety of challenging known and unknown attack protocols. The results reveal interesting trends and can act as pointers for sensor selection for safety-critical presentation attack detection systems. The source codes and protocols to reproduce the results are made available publicly making it possible to extend this work to other architectures. △ Less

Submitted 21 February, 2022; originally announced February 2022.

Comments: 16 pages, 11 images

The High-Quality Wide Multi-Channel Attack (HQ-WMCA) database

Authors: Zohreh Mostaani, Anjith George, Guillaume Heusch, David Geissbuhler, Sebastien Marcel

Abstract: The High-Quality Wide Multi-Channel Attack database (HQ-WMCA) database extends the previous Wide Multi-Channel Attack database(WMCA), with more channels including color, depth, thermal, infrared (spectra), and short-wave infrared (spectra), and also a wide variety of attacks. The High-Quality Wide Multi-Channel Attack database (HQ-WMCA) database extends the previous Wide Multi-Channel Attack database(WMCA), with more channels including color, depth, thermal, infrared (spectra), and short-wave infrared (spectra), and also a wide variety of attacks. △ Less

Submitted 21 September, 2020; originally announced September 2020.

Deep Models and Shortwave Infrared Information to Detect Face Presentation Attacks

Authors: Guillaume Heusch, Anjith George, David Geissbuhler, Zohreh Mostaani, Sebastien Marcel

Abstract: 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… ▽ More 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 bonafide 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. △ Less

Submitted 22 July, 2020; originally announced July 2020.

Multispectral Biometrics System Framework: Application to Presentation Attack Detection

Authors: Leonidas Spinoulas, Mohamed Hussein, David Geissbühler, Joe Mathai, Oswin G. Almeida, Guillaume Clivaz, Sébastien Marcel, Wael AbdAlmageed

Abstract: In this work, we present a general framework for building a biometrics system capable of capturing multispectral data from a series of sensors synchronized with active illumination sources. The framework unifies the system design for different biometric modalities and its realization on face, finger and iris data is described in detail. To the best of our knowledge, the presented design is the fir… ▽ More In this work, we present a general framework for building a biometrics system capable of capturing multispectral data from a series of sensors synchronized with active illumination sources. The framework unifies the system design for different biometric modalities and its realization on face, finger and iris data is described in detail. To the best of our knowledge, the presented design is the first to employ such a diverse set of electromagnetic spectrum bands, ranging from visible to long-wave-infrared wavelengths, and is capable of acquiring large volumes of data in seconds. Having performed a series of data collections, we run a comprehensive analysis on the captured data using a deep-learning classifier for presentation attack detection. Our study follows a data-centric approach attempting to highlight the strengths and weaknesses of each spectral band at distinguishing live from fake samples. △ Less

Submitted 12 June, 2020; originally announced June 2020.

En route to a multi-model scheme for clinker comminution with chemical grinding aids

Authors: R. K. Mishra, D. Geissbuhler, H. A. Carmona, F. K. Wittel, M. L. Sawley, M. Weibel, E. Gallucci, H. J. Herrmann, H. Heinz, R. J. Flatt

Abstract: We present a multi-model simulation approach, targeted at understanding the behavior of comminution and the effect of grinding aids (GAs) in industrial cement mills. On the atomistic scale we use Molecular Dynamics (MD) simulations with validated force field models to quantify elastic and structural properties, cleavage energies as well as the organic interactions with mineral surfaces. Simulation… ▽ More We present a multi-model simulation approach, targeted at understanding the behavior of comminution and the effect of grinding aids (GAs) in industrial cement mills. On the atomistic scale we use Molecular Dynamics (MD) simulations with validated force field models to quantify elastic and structural properties, cleavage energies as well as the organic interactions with mineral surfaces. Simulations based on the Discrete Element Method (DEM) are used to integrate the information gained from MD simulations into the clinker particle behavior at larger scales. Computed impact energy distributions from DEM mill simulations can serve as a link between large-scale industrial and laboratory sized mills. They also provide the required input for particle impact fragmentation models. Such a multi-scale, multi-model methodology paves the way for a structured approach to the design of chemical additives aimed at improving mill performance. △ Less

Submitted 30 July, 2015; originally announced July 2015.

Comments: 9 pages in Advances in Applied Ceramics 2015

Exploring Double Field Theory

Authors: David Geissbuhler, Diego Marques, Carmen Nunez, Victor Penas

Abstract: We present a flux formulation of Double Field Theory, in which geometric and non-geometric fluxes are dynamical and field-dependent. Gauge consistency imposes a set of quadratic constraints on the dynamical fluxes, which can be solved by truly double configurations. The constraints are related to generalized Bianchi Identities for (non-)geometric fluxes in the double space, sourced by (exotic) bra… ▽ More We present a flux formulation of Double Field Theory, in which geometric and non-geometric fluxes are dynamical and field-dependent. Gauge consistency imposes a set of quadratic constraints on the dynamical fluxes, which can be solved by truly double configurations. The constraints are related to generalized Bianchi Identities for (non-)geometric fluxes in the double space, sourced by (exotic) branes. Following previous constructions, we then obtain generalized connections, torsion and curvatures compatible with the consistency conditions. The strong constraint-violating terms needed to make contact with gauged supergravities containing duality orbits of non-geometric fluxes, systematically arise in this formulation. △ Less

Submitted 27 June, 2013; v1 submitted 4 April, 2013; originally announced April 2013.

Comments: To appear in JHEP

Holographic Renormalization for z=2 Lifshitz Space-Times from AdS

Authors: Wissam Chemissany, David Geissbühler, Jelle Hartong, Blaise Rollier

Abstract: Lifshitz space-times with critical exponent z=2 can be obtained by dimensional reduction of Schroedinger space-times with critical exponent z=0. The latter space-times are asymptotically AdS solutions of AdS gravity coupled to an axion-dilaton system and can be uplifted to solutions of type IIB supergravity. This basic observation is used to perform holographic renormalization for 4-dimensional as… ▽ More Lifshitz space-times with critical exponent z=2 can be obtained by dimensional reduction of Schroedinger space-times with critical exponent z=0. The latter space-times are asymptotically AdS solutions of AdS gravity coupled to an axion-dilaton system and can be uplifted to solutions of type IIB supergravity. This basic observation is used to perform holographic renormalization for 4-dimensional asymptotically z=2 locally Lifshitz space-times by Scherk-Schwarz dimensional reduction of the corresponding problem of holographic renormalization for 5-dimensional asymptotically locally AdS space-times coupled to an axion-dilaton system. We can thus define and characterize a 4-dimensional asymptotically locally z=2 Lifshitz space-time in terms of 5-dimensional AdS boundary data. In this setup the 4-dimensional structure of the Fefferman-Graham expansion and the structure of the counterterm action, including the scale anomaly, will be discussed. We find that for asymptotically locally z=2 Lifshitz space-times obtained in this way there are two anomalies each with their own associated nonzero central charge. Both anomalies follow from the Scherk--Schwarz dimensional reduction of the 5-dimensional conformal anomaly of AdS gravity coupled to an axion-dilaton system. Together they make up an action that is of the Horava-Lifshitz type with nonzero potential term for z=2 conformal gravity. △ Less

Submitted 21 August, 2012; v1 submitted 25 May, 2012; originally announced May 2012.

Comments: 32 pages, v2: modified discussion of the central charges

Double Field Theory and N=4 Gauged Supergravity

Authors: David Geissbühler

Abstract: Double Field Theory describes the NS-NS sector of string theory and lives on a doubled spacetime. The theory has a local gauge symmetry generated by a generalization of the Lie derivative for doubled coordinates. For the action to be invariant under this symmetry, a differential constraint is imposed on the fields and gauge parameters, reducing their possible dependence in the doubled coordinates.… ▽ More Double Field Theory describes the NS-NS sector of string theory and lives on a doubled spacetime. The theory has a local gauge symmetry generated by a generalization of the Lie derivative for doubled coordinates. For the action to be invariant under this symmetry, a differential constraint is imposed on the fields and gauge parameters, reducing their possible dependence in the doubled coordinates. We perform a Scherk-Schwarz reduction of Double Field Theory, yielding electric gaugings of half-maximal supergravity in four dimensions when integrability conditions are assumed. The residual symmetries of the compactified theory are mapped with the symmetries of the effective theory and the differential constraints of Double Field Theory are compared with the algebraic conditions on the embedding tensor. It is found that only a weaker form of the differential constraint has to be imposed on background fields to ensure the local gauge symmetry of the reduced action. △ Less

Submitted 5 December, 2011; v1 submitted 20 September, 2011; originally announced September 2011.

Comments: 25 pages; v2: typos corrected, published version

Journal ref: JHEP11(2011)116