Explore the vast range of titles available.

With the widespread use of biometric authentication comes the exploitation of presentation attacks, possibly undermining the effectiveness of these technologies in real-world setups. One example takes place when an impostor, aiming at unlocking someone else's smartphone, deceives the built-in face recognition system by presenting a printed image of the user. In this work, we study the problem of automatically detecting presentation attacks against face authentication methods, considering the use-case of fast device unlocking and hardware constraints of mobile devices. To enrich the understanding of how a purely software-based method can be used to tackle the problem, we present a solely data-driven approach trained with multi-resolution patches and a multi-objective loss function crafted specifically to the problem. We provide a careful analysis that considers several user-disjoint and cross-factor protocols, highlighting some of the problems with current datasets and approaches. Such analysis, besides demonstrating the competitive results yielded by the proposed method, provides a better conceptual understanding of the problem. To further enhance efficacy and discriminability, we propose a method that leverages the available gallery of user data in the device and adapts the method decision-making process to the user's and the device's own characteristics. Finally, we introduce a new presentation-attack dataset tailored to the mobile-device setup, with real-world variations in lighting, including outdoors and low-light sessions, in contrast to existing public datasets.

Learning from fully-unlabeled data is challenging in Multimedia Forensics problems, such as Person Re-Identification and Text Authorship Attribution. Recent self-supervised learning methods have shown to be effective when dealing with fully-unlabeled data in cases where the underlying classes have significant semantic differences, as intra-class distances are substantially lower than inter-class distances. However, this is not the case for forensic applications in which classes have similar semantics and the training and test sets have disjoint identities. General self-supervised learning methods might fail to learn discriminative features in this scenario, thus requiring more robust strategies. We propose a strategy to tackle Person Re-Identification and Text Authorship Attribution by enabling learning from unlabeled data even when samples from different classes are not prominently diverse. We propose a novel ensemble-based clustering strategy whereby clusters derived from different configurations are combined to generate a better grouping for the data samples in a fully-unsupervised way. This strategy allows clusters with different densities and higher variability to emerge, reducing intra-class discrepancies without requiring the burden of finding an optimal configuration per dataset. We also consider different Convolutional Neural Networks for feature extraction and subsequent distance computations between samples. We refine these distances by incorporating context and grouping them to capture complementary information. Our method is robust across both tasks, with different data modalities, and outperforms state-of-the-art methods with a fully-unsupervised solution without any labeling or human intervention.

In unconstrained scenarios, face recognition and person re-identification are subject to distortions such as motion blur, atmospheric turbulence, or upsampling artifacts. To improve robustness in these scenarios, we propose a methodology called Distortion-Adaptive Learned Invariance for Identification (DaliID) models. We contend that distortion augmentations, which degrade image quality, can be successfully leveraged to a greater degree than has been shown in the literature. Aided by an adaptive weighting schedule, a novel distortion augmentation is applied at severe levels during training. This training strategy increases feature-level invariance to distortions and decreases domain shift to unconstrained scenarios. At inference, we use a magnitude-weighted fusion of features from parallel models to retain robustness across the range of images. DaliID models achieve state-of-the-art (SOTA) for both face recognition and person re-identification on seven benchmark datasets, including IJB-S, TinyFace, DeepChange, and MSMT17. Additionally, we provide recaptured evaluation data at a distance of 750+ meters and further validate on real long-distance face imagery.

Person Re-Identification (ReID) across non-overlapping cameras is a challenging task and, for this reason, most works in the prior art rely on supervised feature learning from a labeled dataset to match the same person in different views. However, it demands the time-consuming task of labeling the acquired data, prohibiting its fast deployment, specially in forensic scenarios. Unsupervised Domain Adaptation (UDA) emerges as a promising alternative, as it performs feature-learning adaptation from a model trained on a source to a target domain without identity-label annotation. However, most UDA-based algorithms rely upon a complex loss function with several hyper-parameters, which hinders the generalization to different scenarios. Moreover, as UDA depends on the translation between domains, it is important to select the most reliable data from the unseen domain, thus avoiding error propagation caused by noisy examples on the target data -- an often overlooked problem. In this sense, we propose a novel UDA-based ReID method that optimizes a simple loss function with only one hyper-parameter and that takes advantage of triplets of samples created by a new offline strategy based on the diversity of cameras within a cluster. This new strategy adapts the model and also regularizes it, avoiding overfitting on the target domain. We also introduce a new self-ensembling strategy, in which weights from different iterations are aggregated to create a final model combining knowledge from distinct moments of the adaptation. For evaluation, we consider three well-known deep learning architectures and combine them for final decision-making. The proposed method does not use person re-ranking nor any label on the target domain, and outperforms the state of the art, with a much simpler setup, on the Market to Duke, the challenging Market1501 to MSMT17, and Duke to MSMT17 adaptation scenarios.

Fully-unsupervised Person and Vehicle Re-Identification have received increasing attention due to their broad applicability in surveillance, forensics, event understanding, and smart cities, without requiring any manual annotation. However, most of the prior art has been evaluated in datasets that have just a couple thousand samples. Such small-data setups often allow the use of costly techniques in time and memory footprints, such as Re-Ranking, to improve clustering results. Moreover, some previous work even pre-selects the best clustering hyper-parameters for each dataset, which is unrealistic in a large-scale fully-unsupervised scenario. In this context, this work tackles a more realistic scenario and proposes two strategies to learn from large-scale unlabeled data. The first strategy performs a local neighborhood sampling to reduce the dataset size in each iteration without violating neighborhood relationships. A second strategy leverages a novel Re-Ranking technique, which has a lower time upper bound complexity and reduces the memory complexity from O(n^2) to O(kn) with k << n. To avoid the pre-selection of specific hyper-parameter values for the clustering algorithm, we also present a novel scheduling algorithm that adjusts the density parameter during training, to leverage the diversity of samples and keep the learning robust to noisy labeling. Finally, due to the complementary knowledge learned by different models, we also introduce a co-training strategy that relies upon the permutation of predicted pseudo-labels, among the backbones, with no need for any hyper-parameters or weighting optimization. The proposed methodology outperforms the state-of-the-art methods in well-known benchmarks and in the challenging large-scale Veri-Wild dataset, with a faster and memory-efficient Re-Ranking strategy, and a large-scale, noisy-robust, and ensemble-based learning approach.

Disclosure: G. Grande: None. C. Ceolin: None. B. Vescovi: None. M. Dall’Agnol: None. C. Ziliotto: None. S. Pasqualini: None. G. Petre: None. A. Scala: None. S. Giannini: None. V. Camozzi: None. A. Bertocco: None. G. Sergi: None. A. Ferlin: None. A. Garolla: None. In transgender subjects, worse Bone Mineral Density (BMD) values were observed when compared to cisgender controls before initiating Gender-Affirming Hormonal Therapy (GAHT). The reasons have not yet been fully clarified. In this study we analyzed bone metabolism and body composition in a population of transgender people before initiating GAHT, and their possible correlation with endocrine profile, mental health, and lifestyle habits. Medical data, phosphocalcic metabolism and hormonal parameters, lumbar and femoral bone mineral density, body composition, strength and psychological well-being (by Patient Health Questionnaire-9-PHQ-9 and Perceived Stress Scale-PSS questionnaires), were collected in a sample of 32 transgender people before GAHT initiation and 32 cisgender controls. Assigned Female At Birth (AFAB) transgender people presented worse Z-score values at total hip and neck femoral sites (-0.52±0.94 vs 0.15±0.82, and -0.37±0.79 vs 0.19±0.66, p=0.04 respectively), while in Assigned Male At Birth (AMAB) transgender people also lumbar sites were compromised in comparison to cisgender controls. No significant difference in phospho-calcium metabolism (calcium, phosphate, PTH and vitamin D) or hormonal profile (Testosterone, estradiol, LH, FSH) was found between the transgender and the cisgender group. PHQ-9 (a questionnaire to screen for depression) scores were higher in the transgender population than cisgender controls (9.35±7.41 vs 4.08±3.14, p=0.03 in AFAB and 9.54±5.45 vs 4.46±3.59, p=0.02 in AMAB transgender people). Multiple regression analysis identified PSS (a tool to assess stress levels and identify which situations are perceived as stressful) score as an independent predictor of total femur BMD, explaining about 54% of the variance. Transgender people before GAHT have worse bone health than cisgender subjects. In particular, BMD was reduced at the hip and in AMAB even at lumbar site. The reduction in bone mass before GAHT seems independent by hormonal and phospho-calcic profile and associated mostly with psychological status. Further study will clarify the specific pathogenetic mechanisms involved in bone growth in transgender people. However, according to our evidences, the psychological stress might influence negatively lifestyle habits therefore being involved in reduced bone health. Early lifestyle and psychological interventions might so be beneficial also for bone health in the transgender population. Presentation: Saturday, June 17, 2023