Search Results (549)

Biometric recognition of human gait is a promising, non-invasive method for the identification of people that does not require their engagement. Existing solutions mainly focus on the identification effectiveness under laboratory conditions, frequently overlooking factors that disrupt the gait of test subjects. The present work considers the issue of identifying a person on the basis of ground reaction forces in cases where their gait is disrupted through asymmetric loading. This paper proposes a solution based on ensemble classifiers utilizing various types of deep neural networks as base classifiers. To further increase the ability to generalize base models, data augmentation was used. The proposed solution was tested on a sample of 215 people (7351 gait cycles) and two strategies for combining classifier decisions. The accuracy results obtained, ranging between 99.8, 98.55, and 98.85% correct recognitions depending on the scenario analyzed, are very good and significantly exceed other methods presented in the literature to date. Full article

Finger vein recognition has emerged as a highly robust and intrinsically stable biometric technology, demonstrating great potential in identity authentication and intelligent security applications. However, conventional methods still suffer from constraints in feature representation and computational efficiency, particularly under challenging conditions such as illumination variation, pose deviation, and noise interference. To address these challenges, this study presents an efficient finger vein recognition approach based on a lightweight convolutional neural network (LCNN) architecture. The proposed framework integrates a multi-stage image preprocessing pipeline for automatic vein region detection, advanced denoising, and refined texture enhancement, which is subsequently followed by compact feature modeling within a lightweight deep network. Extensive experiments on the public Shandong University Machine Learning and Applications-Homologous Multi-Modal Traits (SDUMLA-HMT) dataset and a self-acquired Laboratory Finger-Vein (Lab-Vein) dataset validate the superiority of the proposed method, achieving recognition accuracies of 97.1% and 98.3%, respectively, surpassing existing benchmark models. Moreover, the model demonstrates notable reductions in parameter complexity and computational cost, achieving an average inference time of only 12.6 ms, which confirms its strong real-time capability and suitability for embedded deployment. Overall, the proposed approach attains a desirable trade-off between accuracy and efficiency, offering meaningful implications for the advancement of lightweight biometric recognition systems. Full article

Profile images from social networks are a valuable source of data for AI analytics, but they contain biometric identifiers that pose serious privacy risks. The current face anonymization techniques often destroy semantic information, and generative de-identification methods are vulnerable to re-identification attacks. In this paper, we propose a template-driven multimodal face pseudonymization framework that allows for the privacy-preserving analysis of facial image data while retaining analytically relevant attributes. Our approach uses a FaceNet-based CelebA attribute classifier to extract fine-grained facial attributes and a DeepFace model to extract high-level demographic attributes. Rather than relying on stochastic large language models, we introduce deterministic template-based attribute-to-text conversion to ensure consistency and reproducibility and prevent unintended attribute hallucination. The resulting textual description serves as the sole conditioning input for Janus-Pro, a multimodal text-to-image generation model that synthesizes realistic yet non-identifiable face images. We evaluate our method on the CelebA dataset under a strong adversarial threat model, employing state-of-the-art face recognition systems to assess re-identification and linkability attacks. Our results demonstrate a substantial reduction in identity leakage while preserving semantic attributes. Full article

This paper presents the design of a hardware–software system for non-invasive automated monitoring of feeding behavior in livestock with biometric identification of individual animals. Neural network models for animal identification from images and individual recognition have been developed and trained. A solution is proposed to address the challenge of acquiring a sufficient number of personalized animal images for training the identification neural network. A transfer learning approach is introduced for pig identification, where the network is first trained on a large-scale dataset of more than three million human face images obtained from open sources and subsequently fine-tuned by training the upper layers on a significantly smaller dataset consisting of 5610 pig face images. Experimental results demonstrated the high effectiveness of the system: the Top-1 identification accuracy reached 95.1%, while the ROC AUC in open-set recognition tasks achieved 0.95. The processing time per frame on an NVIDIA RTX 4090 GPU was 1.4 ms (724 FPS). Full article

Wrist-worn devices enable new paradigms of implicit and continuous user authentication; however, identifying biometric modalities that combine reliability with practical integrability remains challenging. Inner-wrist skin texture represents a relatively unexplored biometric characteristic that may be acquired unobtrusively using commodity hardware. This study evaluates biometric verification based on inner-wrist skin texture using an off-the-shelf capacitive fingerprint sensor and an unmodified, manufacturer-provided fingerprint verification algorithm. Two experiments were conducted. Experiment 1 assessed baseline verification performance under controlled acquisition conditions in a cohort of 33 participants (21 male, 12 female; mean age 30.0 ± 16.9 years, range 10–71 years), yielding 1768 genuine authentication trials. Experiment 2 examined the effect of wrist posture variation under controlled flexion in a separate cohort of 15 participants (11 male, 4 female; mean age 30.9 years, range 18–49 years), with 3900 authentication trials recorded. Across 86,897 impostor comparisons in Experiment 1, no false acceptances were observed, corresponding to a conservative upper bound on the false acceptance rate of 6.7 × 10⁻⁵ at the 99.7% confidence level, while the false rejection rate was approximately 2.93%. In Experiment 2, the overall false rejection rate increased to 3.52%, with no clear monotonic relationship between wrist angle and verification performance within the tested range. The results demonstrate that inner-wrist skin texture can be captured and matched using fingerprint-oriented sensing and matching technology under controlled conditions, providing an experimental baseline for this biometric modality. At the same time, the use of a closed matching algorithm and a sensor designed for fingerprints limits interpretability and generalization. These findings motivate further investigation using dedicated recognition methods, larger sensing areas, and extended evaluation protocols tailored specifically to wrist skin print biometrics. Full article

Background/Objectives: Safe delivery and correct identification of newborns are critical aspects of healthcare systems globally. The accreditation of healthcare and standards regulation significantly promotes the adoption of modern technologies to address risks related to infant abduction and misidentification. The effectiveness and extent of these mandates vary across settings and countries. Therefore, this study aims to map and explore modern technologies used for safe newborn delivery and correct identification aligned with healthcare accreditation and regulatory frameworks. Methods: This review adheres to the Preferred Reporting Items for Systematic Review and Meta-Analysis extension for scoping reviews (PRISMA-ScR) guidelines. The Problem, Intervention, Comparison, and Outcome (PICO) framework was employed to facilitate the development of the research question. This study examined studies reporting technologies such as radio frequency identification (RFID), biometric identification, and real-time monitoring across healthcare settings for infant protection through the Normalization Process Theory (NPT). Among three databases and search engines (PubMed, Google Scholar, and Web of Science). The risk of bias for each study was assessed using the AACODS Checklist, SQUIRE 2.0 Checklist, TIDieR Checklist, and JBI tools. Results: Out of 8753 records, only 27 reports were eligible to be included in this review. The most frequently reported technologies were RFID systems (11 studies, 37.9%) and biometric systems such as footprint and facial recognition (6 studies, 20.7%). Despite strong technological potential, many healthcare institutions struggled with the adoption of infant protection technologies. Accreditation systems among the high-resource settings actively mandate advanced technologies and support the integration of staff training and simulation drills. Comparably, middle- and low-income regions usually face challenges related to regulatory enforcement, infrastructure, staff readiness, and limited adoption of modern technologies. Conclusions: Accreditation and standards development are critical catalysts for the adoption of modern infant protection technology. Standards must be comprehensible, adaptable, and supported by investment in human resources and infrastructure. Future regulation must focus on strengthening enforcement, continuous quality improvement, and capacity building to achieve sustainable protection across the world. Full article

Fingerprint recognition systems have relied on fragile workflows based on minutiae extraction, which suffer from significant performance losses under real-world conditions such as sensor diversity and low image quality. This study introduces a fully minutiae-free fingerprint recognition framework based on self-supervised Vision Transformers. A systematic evaluation of multiple DINOv2 model variants is conducted, and the proposed system ultimately adopts the DINOv2-Base Vision Transformer as the primary configuration, as it offers the best generalization performance trade-off under conditions of limited fingerprint data. Larger variants are additionally analyzed to assess scalability and capacity limits. The DINOv2 pretrained network is fine-tuned using self-supervised domain adaptation on 64,801 fingerprint images, eliminating all classical enhancement, binarization, and minutiae extraction steps. Unlike the single-sensor protocols commonly used in the literature, the proposed approach is extensively evaluated in a heterogeneous testbed with a wide range of sensors, qualities, and acquisition methods, including 1631 unique fingers from 12 datasets. The achieved EER of 5.56% under these challenging conditions demonstrates clear cross-sensor superiority over traditional systems such as VeriFinger (26.90%) and SourceAFIS (41.95%) on the same testbed. A systematic comparison of different model capacities shows that moderate-scale ViT models provide optimal generalization under limited-data conditions. Explainability analyses indicate that the attention maps of the model trained without any minutiae information exhibit meaningful overlap with classical structural regions (IoU = 0.41 ± 0.07). Openly sharing the full implementation and evaluation infrastructure makes the study reproducible and provides a standardized benchmark for future research. Full article

Periocular recognition is essential when full-face images cannot be used because of occlusion, privacy constraints, or sensor limitations, yet in many deployments, only periocular images are available at run time, while richer evidence, such as archival face photos and textual metadata, exists offline. This mismatch makes it hard to deploy conventional multimodal fusion. This motivates the notion of conditional biometrics, where auxiliary modalities are used only during training to learn stronger periocular representations while keeping deployment strictly periocular-only. In this paper, we propose Multi-Anchor Conditional Periocular Embedding (MACPE), which maps periocular, facial, and textual features into a shared anchor-conditioned space via a learnable anchor bank that preserves periocular micro-textures while aligning higher-level semantics. Training combines identity classification losses on periocular and face branches with a symmetric InfoNCE loss over anchors and a pulling regularizer that jointly aligns periocular, facial, and textual embeddings without collapsing into face-dominated solutions; captions generated by a vision language model provide complementary semantic supervision. At deployment, only the periocular encoder is used. Experiments across five periocular datasets show that MACPE consistently improves Rank-1 identification and reduces EER at a fixed FAR compared with periocular-only baselines and alternative conditioning methods. Ablation studies verify the contributions of anchor-conditioned embeddings, textual supervision, and the proposed loss design. Full article

Self-supervised learning (SSL) models have achieved remarkable success in speaker verification tasks, yet their robustness to real-world audio degradation remains insufficiently characterized. This study presents a comprehensive analysis of how audio quality degradation affects three prominent SSL-based speaker verification systems (WavLM, Wav2Vec2, and HuBERT) across three diverse datasets: TIMIT, CHiME-6, and Common Voice. We systematically applied 21 degradation conditions spanning noise contamination (SNR levels from 0 to 20 dB), reverberation (RT60 from 0.3 to 1.0 s), and codec compression (various bit rates), then measured both objective audio quality metrics (PESQ, STOI, SNR, SegSNR, fwSNRseg, jitter, shimmer, HNR) and speaker verification performance metrics (EER, AUC-ROC, d-prime, minDCF). At the condition level, multiple regression with all eight quality metrics explained up to 80% of the variance in minDCF for HuBERT and 78% for WavLM, but only 35% for Wav2Vec2; EER predictability was lower (69%, 67%, and 28%, respectively). PESQ was the strongest single predictor for WavLM and HuBERT, while Shimmer showed the highest single-metric correlation for Wav2Vec2; fwSNRseg yielded the top single-metric R² for WavLM, and PESQ for HuBERT and Wav2Vec2 (with much smaller gains for Wav2Vec2). WavLM and HuBERT exhibited more predictable quality-performance relationships compared to Wav2Vec2. These findings establish quantitative relationships between measurable audio quality and speaker verification accuracy at the condition level, though substantial within-condition variability limits utterance-level prediction accuracy. Full article

Near-infrared organic photodetectors (NIR-OPDs) are emerging as versatile platforms for flexible and low-cost optical sensing, yet achieving high-performance in the NIR region remains difficult remains challenging due to intrinsic trade-offs at both the material and device levels, due to the inherent balance required among bandgap narrowing, exciton dissociation, charge transport, and dark-current suppression. This review provides a concise overview of OPD operating mechanisms and the performance metrics governing sensitivity and noise. We highlight recent molecular-engineering strategies—core fluorination, asymmetric π-bridge design, fused-ring rigidification, and polymer backbone/side-chain tuning—that effectively enhance intermolecular ordering, reduce energetic disorder, and extend NIR absorption. Progress in all-polymer detectors and ambipolar phototransistors further demonstrates improved stability and broadened detection capability. Additionally, emerging applications, including NIR communication, biosignal monitoring, flexible imaging, and biometric recognition, showcase the expanding utility of NIR-OPDs. Remaining challenges include pushing detection beyond 1200 nm, simplifying synthesis, and improving long-term stability. Overall, advances in low-bandgap molecular design and device engineering continue to accelerate the practical adoption of NIR-OPDs. Full article

Individual animal identification is a cornerstone of animal welfare practices and is of crucial importance for food safety and the protection of humans from zoonotic diseases. It is also a key prerequisite for enabling automated processes in modern dairy farming. With newly emerging technologies, visual animal identification based on machine learning offers a more efficient and non-invasive method with high automation potential, accuracy, and practical applicability. However, a common challenge is the limited variability of training datasets, as images are typically captured in controlled environments with uniform backgrounds and fixed poses. This study investigates the impact of foreground segmentation and background removal on the performance of convolutional neural networks (CNNs) for cow identification. A dataset was created in which training images of dairy cows exhibited low variability in pose and background for each individual, whereas the test dataset introduced significant variation in both pose and environment. Both a fine-tuned CNN backbone and a model trained from scratch were evaluated using images with and without background information. The results demonstrate that although training on segmented foregrounds extracts intrinsic biometric features, background cues carry more information for individual recognition. Full article

This pilot study evaluated the visionMC system, a low-cost artificial intelligence system integrating HOG-based facial recognition and voice notifications, for workflow optimization in a family medicine practice. Implemented on a Raspberry Pi 4, the system was tested over two weeks with 50 patients. It achieved 85% recognition accuracy and an average detection time of 3.4 s. Compared with baseline, patient waiting times showed a substantial reduction in waiting time and administrative workload, and the administrative workload decreased by 5–7 min per patient. A satisfaction survey (N = 35) indicated high acceptance, with all scores above 4.5/5, particularly for usefulness and waiting time reduction. These results suggest that visionMC can improve efficiency and enhance patient experience with minimal financial and technical requirements. Larger multicenter studies are warranted to confirm scalability and generalizability. visionMC demonstrates that effective AI integration in small practices is feasible with minimal resources, supporting scalable digital health transformation. Beyond biometric identification, the system’s primary contribution is streamlining practice management by instantly displaying the arriving patient and enabling rapid chart preparation. Personalized greetings enhance patient experience, while email alerts on motion events provide a secondary security benefit. These combined effects drove the observed reductions in waiting and administrative times. Full article

While face recognition systems are increasingly deployed in critical domains, they remain vulnerable to presentation attacks and exhibit significant demographic bias, particularly affecting African populations. This paper presents a fairness-aware Presentation Attack Detection (PAD) system using Local Binary Patterns (LBPs) with novel ethnicity-aware processing techniques specifically designed for African contexts. Our approach introduces three key technical innovations: (1) adaptive preprocessing with differentiated Contrast-Limited Adaptive Histogram Equalization (CLAHE) parameters and gamma correction optimized for different skin tones, (2) group-specific decision threshold optimization using Equal Error Rate (EER) minimization for each ethnic group, and (3) three novel statistical methods for PAD fairness evaluation such as Coefficient of Variation analysis, McNemar’s significance testing, and bootstrap confidence intervals representing the first application of these techniques in Presentation Attack Detection. Comprehensive evaluation on the Chinese Academy of Sciences Institute of Automation-SURF Cross-ethnicity Face Anti-spoofing dataset (CASIA-SURF CeFA) dataset demonstrates significant bias reduction achievements: a 75.6% reduction in the accuracy gap between African and East Asian subjects (from 3.07% to 0.75%), elimination of statistically significant bias across all ethnic group comparisons, and strong overall performance, with 95.12% accuracy and 98.55% AUC. Our work establishes a comprehensive methodology for measuring and mitigating demographic bias in PAD systems while maintaining security effectiveness, contributing both technical innovations and statistical frameworks for inclusive biometric security research. Full article

In modern industrial environments, camera-based monitoring is essential for workflow optimization, safety, and process control, yet it raises significant privacy concerns when people are recorded. Realistic full-body anonymization offers a potential solution by obscuring visual identity while preserving information needed for automated analysis. Whether such methods also conceal biometric traits from human pose and gait remains uncertain, although these biomarkers enable person identification without appearance cues. This study investigates the impact of full-body anonymization on gait-related identity recognition using DeepPrivacy2 and a custom CCTV-like industrial dataset comprising original and anonymized sequences. This study provides the first systematic evaluation of whether pose-preserving anonymization disrupts identity-relevant gait characteristics. The analysis quantifies keypoint shifts introduced by anonymization, examines their influence on downstream gait-based person identification, and tests cross-domain linkability between original and anonymized recordings. Identification accuracy, domain transfer between data types, and distortions in derived pose keypoints are measured to assess anonymization effects while retaining operational utility. Findings show that anonymization removes appearance but leaves gait identity largely intact, indicating that pose-driven anonymization is insufficient for privacy protection. Effective privacy requires anonymization strategies that explicitly target gait characteristics or incorporate domain-adaptation mechanisms. Full article

Biometric authentication is widely regarded as more secure and reliable than conventional approaches like passwords and PINs. Nonetheless, many current systems rely on active user participation, such as fingerprint scanning or facial recognition, which can disrupt tasks, increase the likelihood of errors, and raise privacy concerns. To address these challenges, this study introduces a continuous, seamless authentication framework that utilizes vital signs for passive identity verification across various activities, including resting, walking, and running. The framework analyzes physiological indicators such as Heart Rate (HR), Heart Rate Variability (HRV), Skin Temperature, Peripheral Oxygen Saturation (${\mathrm{SpO}}_2$), and Breathing Rate to provide zero-effort authentication without requiring user intervention. Multiple machine learning algorithms, including Decision Tree, Random Forest, XGBoost, Gradient Boosting, and K-Nearest Neighbors, were implemented and compared to identify the most effective predictive model. The methodology involved data collection, preprocessing, model construction, evaluation, and comparison. Experimental results revealed that the XGBoost Classifier achieved the highest accuracy at 96%. Overall, the proposed framework demonstrates strong reliability, scalability, adaptability, and flexibility, making it suitable for practical deployment. By continuously verifying identity without interrupting user activity, it improves both security and usability, offering a modern and convenient alternative to traditional authentication methods. Full article