Multi-face tracking is a challenging problem, as low-quality face images (e.g., partially occluded faces, lateral faces, motion blur, etc.) are prevalent. A multi-face tracker called FaceQSORT mitigates this problem by combining two different types of features (i.e., biometric face features and generic appearance features) extracted from the same image (face) patch. In particular, the appearance features provide a general visual representation of the image (face) patch and compensate for limited biometric information. To extract these appearance features, a generic image classifier is utilized. In this work, the selection of the appearance feature model (image classifier) and the biometric feature model (face recognition model) is evaluated empirically. For this purpose, the following questions are investigated: (i) How crucial is the choice of appearance model? (ii) How does the choice of appearance model affect the inter-person ID switches? To answer these questions, a comprehensive experimental evaluation is conducted that includes multiple tracking parameter settings and different combinations of biometric and appearance model pairs.
This paper investigates the impact of multi-sample fusion at enrolment on the performance and security of biometric cryptosystems, specifically focusing on the fuzzy vault scheme. Biometric template protection is critical for safeguarding sensitive biometric data, and we propose enhancing robustness by averaging multiple biometric samples from the same instance of a modality during the enrolment phase. This approach aims to reduce intra-class variability and stabilize feature representations, thereby improving recognition accuracy without significantly impacting usability. We evaluate a fuzzy vault system based on deep learning-based fingerprint embeddings, using a comprehensive fingerprint dataset (MCYT330) with varying sample qualities. Our experimental results demonstrate that fusing multiple enrolment samples significantly shifts mated score distributions, leading to notable improvements in recognition rates and security in both unprotected and fuzzy vault systems. This work confirms the practical benefits of multi-sample fusion in developing more reliable and secure privacy-preserving biometric systems.
Iris segmentation is an important step in determining downstream iris recognition performance. Most iris segmentation systems fail to reliably segment the iris images at extreme off-angles that go beyond 30 degrees. Obtaining a manual ground truth for different extreme angles is both a costly and a time-consuming undertaking. In this paper, our aim is to automate this iris segmentation process for different extreme gaze angles, while minimizing the iris recognition performance loss. To address this, we propose an Iris segmentation method using a novel distance-based loss function and a residual-based ellipse fitting method to overcome these obstacles. The novel method shows an overall best performance of 0.9809 Area Under Curve (AUC) and an equal error rate (EER) of 0.0628, reducing the gap to the performance using ideal manual segmentation to 1.88% in terms of AUC and 0.056% in terms of EER.
Tattoo recognition is a critical task in forensics; however, existing approaches often face limitations under real-world conditions such as occlusion, distortion, and low image quality. This paper presents a novel and robust tattoo retrieval framework that combines the global semantic representation capabilities of OmniGlue with the geometric precision of LightGlue. Notably, the proposed framework is entirely based on foundation models and operates in a training-free manner, enabling rapid deployment without the need for dataset-specific fine-tuning. Extensive experiments conducted on the challenging WebTattoo dataset demonstrate the effectiveness of the approach. In closed-set evaluation, the framework achieves a rank-1 identification rate of 90.91%, significantly outperforming standalone models as well as the state-of-the-art TattTRN. In open-set identification, the method attains an Equal Error Rate of 11.56%, outper-forming equally the state of the art, i.e., TattTRN (20.75%). These results highlight the advantages of integrating complementary foundation models through mechanisms, establishing a new benchmark for tattoo recognition in both controlled and unconstrained environments.
This paper presents a pilot study on the feasibility of using a selected commercial electronic fingerprint biometric capture subsystem for post mortem biometric recognition. The study investigates the effectiveness of contact and contactless devices in capturing fingerprints from deceased individuals within 48 hours after death. The environmental conditions, legal and medical implications, and post mortem changes that affect skin integrity are examined. Fingerprint quality and interoperability were evaluated using the NFIQ 2 and IDkit tools. The results show that while traditional dactyloscopic cards offer lower image quality for algorithmic comparison, some modern digital devices provide promising results for post mortem fingerprint acquisition and comparison.
