Logo des Repositoriums

Auflistung nach Autor:in "Raja, Kiran"

1 - 10 von 136
  • Konferenzbeitrag
    3D Face Recognition For Cows
    (BIOSIG 2020 - Proceedings of the 19th International Conference of the Biometrics Special Interest Group, 2020) Yeleshetty, Deepak; Spreeuwers, Luuk; Li, Yan
    This paper presents a method to recognize cows using their 3D face point clouds. Face is chosen because of the rigid structure of the skull compared to other parts. The 3D face point clouds are acquired using a newly designed dual 3D camera setup. After registering the 3D faces to a specific pose, the cow’s ID is determined by running Iterative Closest Point (ICP) method on the probe against all the point clouds in the gallery. The root mean square error (RMSE) between the ICP correspondences is used to identify the cows. The smaller the RMSE, the more likely that the cow is from the same class. In a closed set of 32 cows with 5 point clouds per cow in the gallery, the ICP recognition demonstrates an almost perfect identification rate of 99.53%.
  • Konferenzbeitrag
    Action-Independent Generalized Behavioral Identity Descriptors for Look-alike Recognition in Videos
    (BIOSIG 2020 - Proceedings of the 19th International Conference of the Biometrics Special Interest Group, 2020) Khodabakhsh, Ali; Loiselle, Hugo
    There is a long history of exploitation of the visual similarity of look-alikes for fraud and deception. The visual similarity along with the application of physical and digital cosmetics greatly challenges the recognition ability of average humans. Face recognition systems are not an exception in this regard and are vulnerable to such similarities. In contrast to physiological face recognition, behavioral face recognition is often overlooked due to the outstanding success of the former. However, the behavior of a person can provide an additional source of discriminative information with regards to the identity of individuals when physiological attributes are not reliable. In this study, we propose a novel biometric recognition system based only on facial behavior for the differentiation of look-alikes in unconstrained recording conditions. To this end, we organized a dataset of 85;656 utterances from 1000 look-alike pairs based on videos collected from the wild, large enough for the development of deep learning solutions. Our selection criteria assert that for these collected videos, both state-of-the-art biometric systems and human judgment fail in recognition. Furthermore, to utilize the advantage of large-scale data, we introduce a novel action-independent biometric recognition system that was trained using triplet-loss to create generalized behavioral identity embeddings. We achieve look-alike recognition equal-error-rate of 7:93% with sole reliance on the behavior descriptors extracted from facial landmark movements. The proposed method can have applications in face recognition as well as presentation attack detection and Deepfake detection.
  • Konferenzbeitrag
    Analysis of Minutiae Quality for Improved Workload Reduction in Fingerprint Identification
    (BIOSIG 2022, 2022) Daile Osorio-Roig, Tim Rohwedder
    The workload of biometric identification in large fingerprint databases poses a challenging problem. Efficient schemes for biometric workload reduction are a topic of ongoing research. Some of the state-of-the art approaches rely on triangles of minutia points generated by Delaunay triangulation, which are then used for indexing. In this paper, we investigate how quality estimation at the minutia level can improve the performance of such algorithms and hence the system workload. In order to reduce the number of spurious and missing minutiae, we analyse the impact of selecting minutiae points based on their qualities. This, in turn, can significantly distort the triangulation. In addition, we consider the usefulness of the average minutia quality as an additional criteria of the minutia triangles for indexing. Our results show that both strategies lead to a significant reduction in biometric workload compared to a baseline solution (i.e. exhaustive search) – down to 36% on average.
  • Konferenzbeitrag
    An anomaly detection approach for backdoored neural networks: face recognition as a case study
    (BIOSIG 2022, 2022) Alexander Unnervik and Sébastien Marcel
    Backdoor attacks allow an attacker to embed functionality jeopardizing proper behavior of any algorithm, machine learning or not. This hidden functionality can remain inactive for normal use of the algorithm until activated by the attacker. Given how stealthy backdoor attacks are, consequences of these backdoors could be disastrous if such networks were to be deployed for applications as critical as border or access control. In this paper, we propose a novel backdoored network detection method based on the principle of anomaly detection, involving access to the clean part of the training data and the trained network.We highlight its promising potential when considering various triggers, locations and identity pairs, without the need to make any assumptions on the nature of the backdoor and its setup. We test our method on a novel dataset of backdoored networks and report detectability results with perfect scores.
  • Konferenzbeitrag
    Application of affine-based reconstruction to retinal point patterns
    (BIOSIG 2020 - Proceedings of the 19th International Conference of the Biometrics Special Interest Group, 2020) Sadeghpour, Mahshid; Arakala, Arathi; Davis, Stephen A.; Horadam, Kathy J.
    Inverse biometrics that exploit the information of biometric references from comparison scores can compromise sensitive personal information of the users in biometric recognition systems. One inverse biometric method that has been very successful in regenerating face images applies an affine transformation to model the face recognition algorithm. This method is general and could apply to templates extracted from other biometric characteristics. This research proposes two formats to apply this method to spatial point patterns extracted from retina images and tests its performance on reconstructing such sparse templates. The results show that the quality of the reconstructed retina point pattern templates is lower than would be accepted by the system as mated.
  • Konferenzbeitrag
    Assessing the Human Ability to Recognize Synthetic Speech in Ordinary Conversation
    (BIOSIG 2023, 2023) Daniel Prudký, Anton Firc
    This work assesses the human ability to recognize synthetic speech (deepfake). This paper describes an experiment in which we communicated with respondents using voice messages. We presented the respondents with a cover story about testing the user-friendliness of voice messages while secretly sending them a pre-prepared deepfake recording during the conversation. We examined their reactions, knowledge of deepfakes, or how many could correctly identify which message was deepfake. The results show that none of the respondents reacted in any way to the fraudulent deepfake message, and only one retrospectively admitted to noticing something specific. On the other hand, a voicemail message that contained a deepfake was correctly identified by 83.9% of respondents after revealing the nature of the experiment. Thus, the results show that although the deepfake recording was clearly identifiable among others, no one reacted to it. In summary, we show that the human ability to recognize voice deepfakes is not at a level we can trust. It is very difficult for people to distinguish between real and fake voices, especially if they do not expect them.
  • Konferenzbeitrag
    Assessment of Sensor Ageing-Impact in Air Travelled Fingerprint Capturing Devices
    (BIOSIG 2021 - Proceedings of the 20th International Conference of the Biometrics Special Interest Group, 2021) Kauba, Christof; Kirchgasser, Simon; Jöchl, Robert; Uhl, Andreas
    Biometric recognition performance is affected by many factors, like varying acquisition conditions or ageing related effects, commonly denoted as biometric template ageing. Image sensor ageing, being part of biometric template ageing and a sub-field of image and video forensics, leads to defective pixels due to cosmic radiation, depending on the altitude. So far, image sensor ageing has only been a peripheral target in fingerprint research. We investigate the impact of image sensor ageing on various fingerprint capturing devices, including optical, capacitive and thermal ones. We established a fingerprint ageing dataset utilising 10 capturing devices which travelled on an air-plane for 127 days (to increase the number of developed defects). By evaluating the samples captured prior to their travel and afterwards using several state-of-the-art fingerprint quality metrics as well as minutiae-based fingerprint recognition systems we quantify the effect of image sensor ageing on fingerprint recognition. Furthermore, by employing a defect detection technique we quantify the number of defects developed during that period.
  • Konferenzbeitrag
    Automatic validation of ICAO compliance regarding head coverings: an inclusive approach concerning religious circumstances
    (BIOSIG 2023, 2023) Carla Guerra, João S. Marcos
    This paper contributes with a dataset and an algorithm that automatically verifies the compliance with the ICAO requirements related to the use of head coverings on facial images used on machine-readable travel documents. All the methods found in the literature ignore that some coverings might be accepted because of religious or cultural reasons, and basically only look for the presence of hats/caps. Our approach specifically includes the religious cases and distinguishes the head coverings that might be considered compliant. We built a dataset composed by facial images of 500 identities to accommodate these type of accessories. That data was used to fine-tune and train a classification model based on the YOLOv8 framework and we achieved state of the art results with an accuracy of 99.1% and EER of 5.7%.
  • Konferenzbeitrag
    Benchmarking fixed-length Fingerprint Representations across different Embedding Sizes and Sensor Types
    (BIOSIG 2023, 2023) Tim Rohwedder, Daile Osorio Roig
    Traditional minutiae-based fingerprint representations consist of a variable-length set of minutiae. This necessitates a more complex comparison causing the drawback of high computational cost in one-to-many comparison. Recently, deep neural networks have been proposed to extract fixed-length embeddings from fingerprints. In this paper, we explore to what extent fingerprint texture information contained in such embeddings can be reduced in terms of dimension, while preserving high biometric performance. This is of particular interest, since it would allow to reduce the number of operations incurred at comparisons. We also study the impact in terms of recognition performance of the fingerprint textural information for two sensor types, i.e. optical and capacitive. Furthermore, the impact of rotation and translation of fingerprint images on the extraction of fingerprint embeddings is analysed. Experimental results conducted on a publicly available database reveal an optimal embedding size of 512 feature elements for the texture-based embedding part of fixed-length fingerprint representations. In addition, differences in performance between sensor types can be perceived. The source code of all experiments presented in this paper is publicly available at https://github.com/tim-rohwedder/fixed-length-fingerprint-extractors, so our work can be fully reproduced.
  • Konferenzbeitrag
    Biometric Recognition in a Multi-sample Multi-Subject Facial Image Database: The 1:M:N System Model
    (BIOSIG 2021 - Proceedings of the 20th International Conference of the Biometrics Special Interest Group, 2021) Halfen, DeWayne; Rajaraman, Srinivasan; Wayman, James L.
    Over the last 50 years, biometric recognition has advanced from localized “identity verification” applications [GU77][RY74] to include large-scale systems in which “a determination is made as to the identity of an individual independently of any information supplied by the individual” [GU77]. Models for estimating and expressing system error rates (both false matches and false non-matches) have been largely limited to so-called “1-to-1” and “1-to-N” systems in which each identity is represented by only one enrolled reference [Gr21]. In this paper, we create a highly simplified simulation model for a common current situation in which each known identity record has multiple stored references. We call this the “1:M:N” model and show that both DET and CMC performance depend upon the number of identities and images per identity, not simply the total number of references images, as usually assumed. Although trialed here on very simple decision policies, this model will be extended in future work to more complex decision criteria.