Lecture Notes in Informatics

Permanent URI for this communityhttps://dl.gi.de/handle/20.500.12116/21

Die „GI-Edition: Lecture Notes in Informatics" (LNI) ist eine eigene Veröffentlichungsreihe der GI mit den Strängen: Proceedings, Dissertations, Seminars und Thematics. Alle in den LNI herausgegebenen Bände werden von GI-Gliederungen unterstützt und verantwortet.

Information und Ansprechpartner zur Reihe finden sich auf den Webseiten der GI unter https://gi.de/service/publikationen/lni

News

Authors with most Documents

Schroeder, Ulrik

Lucke, Ulrike

Busch, Christoph

Thomas, Oliver

Theuvsen, Ludwig

Browse

Search Results

1 - 10 of 59

Evaluating Face Image Quality Score Fusion for Modern Deep Learning Models
(Gesellschaft für Informatik e.V., 2022) Schlett, Torsten; Rathgeb, Christian; Tapia, Juan E.; Busch, Christoph; Brömme, Arslan; Damer, Naser; Gomez-Barrero, Marta; Raja, Kiran; Rathgeb, Christian; Sequeira Ana F.; Todisco, Massimiliano; Uhl, Andreas
Face image quality assessment algorithms attempt to estimate the utility of face images for biometric systems, typically face recognition, since the performance of these systems can be limited by the image quality. Hand-crafted quality score fusion has previously been examined for a variety of mostly factor-specific quality assessment algorithms. This paper instead examines score fusion for various recent “monolithic” quality assessment deep learning models. The evaluation methodology is based on Error-versus-Reject-Characteristic partial-Area-Under-Curve values, which are used to quantitatively rank quality assessment configurations in a face recognition context. Mean quality score fusion configurations were found to slightly improve performance on the TinyFace database, while the tested fusion types were ineffective on the LFW database.
Impact of Doppelgängers on Face Recognition: Database and Evaluation
(Gesellschaft für Informatik e.V., 2021) Rathgeb, Christian; Drozdowski, Pawel; Obel, Marcel; Dörsch, André; Stockhardt, Fabian; Haryanto, Nathania E.; Bernardo, Kevin; Busch, Christoph; Brömme, Arslan; Busch, Christoph; Damer, Naser; Dantcheva, Antitza; Gomez-Barrero, Marta; Raja, Kiran; Rathgeb, Christian; Sequeira, Ana; Uhl, Andreas
Look-alikes, a.k.a. doppelgängers, increase the probability of false matches in a facial recognition system, in contrast to random face image pairs selected for non-mated comparison trials. In order to analyse and improve the robustness of automated face recognition, datasets of doppelgänger face image pairs are needed. In this work, we present a new face database consisting of 400 pairs of doppelgänger images. Subsequently, two state-of-the-art face recognition systems are evaluated on said database and other public datasets, including the Disguised Faces in The Wild (DFW) database. It is found that the collected image pairs yield very high similarity scores resulting in a significant increase of false match rates. To facilitate reproducible research and future experiments in this field, the dataset is made available.
Unit-Selection Based Facial Video Manipulation Detection
(Gesellschaft für Informatik e.V., 2020) Nielsen, V; Khodabakhsh, Ali; Busch, Christoph; Brömme, Arslan; Busch, Christoph; Dantcheva, Antitza; Raja, Kiran; Rathgeb, Christian; Uhl, Andreas
Advancements in video synthesis technology have caused major concerns over the authenticity of audio-visual content. A video manipulation method that is often overlooked is inter-frame forgery, in which segments (or units) of an original video are reordered and rejoined while cut-points are covered with transition effects. Subjective tests have shown the susceptibility of viewers in mistaking such content as authentic. In order to support research on the detection of such manipulations, we introduce a large-scale dataset of 1000 morph-cut videos that were generated by automation of the popular video editing software Adobe Premiere Pro. Furthermore, we propose a novel differential detection pipeline and achieve an outstanding frame-level detection accuracy of 95%.
Simulation of Print-Scan Transformations for Face Images based on Conditional Adversarial Networks
(Gesellschaft für Informatik e.V., 2020) Mitkovski, Aleksandar; Merkle, Johannes; Rathgeb, Christian; Tams, Benjamin; Bernardo, Kevin; Haryanto, Nathania E.; Busch, Christoph; Brömme, Arslan; Busch, Christoph; Dantcheva, Antitza; Raja, Kiran; Rathgeb, Christian; Uhl, Andreas
In many countries, printing and scanning of face images is frequently performed as part of the issuance process of electronic travel documents, e.g., ePassports. Image alterations induced by such print-scan transformations may negatively effect the performance of various biometric subsystems, in particular image manipulation detection. Consequently, according training data is needed in order to achieve robustness towards said transformations. However, manual printing and scanning is time-consuming and costly. In this work, we propose a simulation of print-scan transformations for face images based on a Conditional Generative Adversarial Network (cGAN). To this end, subsets of two public face databases are manually printed and scanned using different printer-scanner combinations. A cGAN is then trained to perform an image-to-image translation which simulates the corresponding print-scan transformations. The goodness of simulation is evaluated with respect to image quality, biometric sample quality and performance, as well as human assessment.
Touchless Fingerprint Sample Quality: Prerequisites for the Applicability of NFIQ2.0
(Gesellschaft für Informatik e.V., 2020) Priesnitz, Jannis; Rathgeb, Christian; Buchmann, Nicolas; Busch, Christoph; Brömme, Arslan; Busch, Christoph; Dantcheva, Antitza; Raja, Kiran; Rathgeb, Christian; Uhl, Andreas
The impact of fingerprint sample quality on biometric performance is undisputed. For touch-based fingerprint data, the effectiveness of the NFIQ2.0 quality estimation method is well documented in scientific literature. Due to the increasing use of touchless fingerprint recognition systems a thorough investigation of the usefulness of the NFIQ2.0 for touchless fingerprint data is of interest. In this work, we investigate whether NFIQ2.0 quality scores are predictive of error rates associated with the biometric performance of touchless fingerprint recognition. For this purpose, we propose a touchless fingerprint preprocessing that favours NFIQ2.0 quality estimation which has been designed for touch-based fingerprint data. Comparisons are made between NFIQ2.0 score distributions obtained from touch-based and touchless fingerprint data of the publicly available FVC06, MCYT, PolyU, and ISPFDv1 databases. Further, the predictive power regarding biometric performance is evaluated in terms of Error-versus-Reject Curves (ERCs) using an open source fingerprint recognition system. Under constrained capture conditions NFIQ2.0 is found to be an effective tool for touchless fingerprint quality estimation if an adequate preprocessing is applied.
Fisher Vector Encoding of Dense-BSIF Features for Unknown Face Presentation Attack Detection
(Gesellschaft für Informatik e.V., 2020) González-Soler, Lázaro J.; Gomez-Barrero, Marta; Busch, Christoph; Brömme, Arslan; Busch, Christoph; Dantcheva, Antitza; Raja, Kiran; Rathgeb, Christian; Uhl, Andreas
The task of determining whether a sample stems from a real subject (i.e, it is a bona fide presentation) or it comes from an artificial replica (i.e., it is an attack presentation) is a mandatory requirement for biometric capture devices, which has received a lot of attention in the recent past. Nowadays, most face Presentation Attack Detection (PAD) approaches have reported a good detection performance when they are evaluated on known Presentation Attack Instruments (PAIs) and acquisition conditions, in contrast to more challenging scenarios where unknown attacks are included in the evaluation. For those more realistic scenarios, the existing approaches are in many cases unable to detect unknown PAI species. In this work, we introduce a new feature space based on Fisher vectors, computed from compact Binarised Statistical Image Features (BSIF) histograms, which allows finding semantic feature subsets from known samples in order to enhance the detection of unknown attacks. This new representation, evaluated over three freely available facial databases, shows promising results in the top state-of-the-art: a BPCER100 under 17% together with a AUC over 98% can be achieved in the presence of unknown attacks.
Can Generative Colourisation Help Face Recognition?
(Gesellschaft für Informatik e.V., 2020) Drozdowski, Pawel; Fischer, Daniel; Rathgeb, Christian; Geissler, Julian; Knedlik, Jan; Busch, Christoph; Brömme, Arslan; Busch, Christoph; Dantcheva, Antitza; Raja, Kiran; Rathgeb, Christian; Uhl, Andreas
Generative colourisation methods can be applied to automatically convert greyscale images to realistically looking colour images. In a face recognition system, such techniques might be employed as a pre-processing step in scenarios where either one or both face images to be compared are only available in greyscale format. In an experimental setup which reflects said scenarios, we investigate if generative colourisation can improve face sample utility and overall biometric performance of face recognition. To this end, subsets of the FERET and FRGCv2 face image databases are converted to greyscale and colourised applying two versions of the DeOldify colourisation algorithm. Face sample quality assessment is done using the FaceQnet quality estimator. Biometric performance measurements are conducted for the widely used ArcFace system with its built-in face detector and reported according to standardised metrics. Obtained results indicate that, for the tested systems, the application of generative colourisation does neither improve face image quality nor recognition performance. However, generative colourisation was found to aid face detection and subsequent feature extraction of the used face recognition system which results in a decrease of the overall false reject rate.
A Generalizable Deepfake Detector based on Neural Conditional Distribution Modelling
(Gesellschaft für Informatik e.V., 2020) Khodabakhsh, Ali; Busch, Christoph; Brömme, Arslan; Busch, Christoph; Dantcheva, Antitza; Raja, Kiran; Rathgeb, Christian; Uhl, Andreas
Photo- and video-realistic generation techniques have become a reality following the advent of deep neural networks. Consequently, there are immense concerns regarding the difficulty in differentiating what content is real from what is synthetic. An example of video-realistic generation techniques is the infamous Deepfakes, which exploit the main modality by which humans identify each other. Deepfakes are a category of synthetic face generation methods and are commonly based on generative adversarial networks. In this article, we propose a novel two-step synthetic face image detection method in which general-purpose features are extracted in a first step, trivializing the task of detecting synthetic images. The anomaly detector predicts the conditional probabilities for observing every individual pixel in the image and is trained on pristine data only. The extracted anomaly features demonstrate true generalization capacity across widely different unknown synthesis methods while showing a minimal loss in performance with regard to the detection of known synthetic samples.
Efficiency Analysis of Post-quantum-secure Face Template Protection Schemes based on Homomorphic Encryption
(Gesellschaft für Informatik e.V., 2020) Kolberg, Jascha; Drozdowski, Pawel; Gomez-Barrero, Marta; Rathgeb, Christian; Busch, Christoph; Brömme, Arslan; Busch, Christoph; Dantcheva, Antitza; Raja, Kiran; Rathgeb, Christian; Uhl, Andreas
Since biometric characteristics are not revocable and biometric data is sensitive, privacypreserving methods are essential to operate a biometric recognition system. More precisely, the biometric information protection standard ISO/IEC IS 24745 requires that biometric templates are stored and compared in a secure domain. Using homomorphic encryption (HE), we can ensure permanent protection since mathematical operations on the ciphertexts directly correspond to those on the plaintexts. Thus, HE allows to compute the distance between two protected templates in the encrypted domain without a degradation of biometric performance with respect to the corresponding system. In this paper, we benchmark three post-quantum-secure HE schemes, and thereby show that a face verification in the encrypted domain requires only 50 ms transaction time and a template size of 5.5 KB.
On the Application of Homomorphic Encryption to Face Identification
(Gesellschaft für Informatik e.V., 2019) Drozdowski, Pawel; Buchmann, Nicolas; Rathgeb, Christian; Margraf, Marian; Busch, Christoph; Brömme, Arslan; Busch, Christoph; Dantcheva, Antitza; Rathgeb, Christian; Uhl, Andreas
The data security and privacy of enrolled subjects is a critical requirement expected from biometric systems. This paper addresses said topic in facial biometric identification. In order to fulfil the properties of unlinkability, irreversibility, and renewability of the templates required for biometric template protection schemes, homomorphic encryption is utilised. In addition to achieving the aforementioned objectives, the use of homomorphic encryption ensures that the biometric performance remains completely unaffected by the template protection scheme. The main contributions of this paper are: It proposes an architecture of a system capable of performing biometric identification in the encrypted domain, as well as provides and evaluates an implementation using two existing homomorphic encryption schemes. Furthermore, it discusses the pertinent technical considerations and challenges in this context.