Publications by Guido Borghi

Ensuring that facial images conform to widely adopted quality guidelines is a crucial step in optimizing the document enrollment workflow, which includes the face verification task. In this paper, we focus on the ISO/ICAO standard, which defines the requirements for facial photographs used in official documents, such as passports, ensuring consistency in face quality and thereby improving reliable recognition by both humans and biometric systems. Generally, ISO/ICAO compliance verification is manually performed through a slow, subjective, and non-scalable process, then to address these challenges, we introduce a fully automated system that assesses face compliance directly from the official standard requirements, eliminating dependence on predefined, hand-crafted features and empirically set thresholds. The method integrates a language model with an innovative prompt learning strategy and a contrastive learning paradigm to assess whether a given facial image satisfies specific quality criteria. Experimental evaluations demonstrate that our method achieves competitive accuracy compared to both academic and commercial baselines. By facilitating the integration and maintenance of compliance regulations, the proposed framework offers a practical, scalable, and regulation-centric solution for automated image quality verification. All code and models are publicly available1.

Technologies to enable safe and effective collaboration and coexistence between humans and robots have gained significant importance in the last few years. A critical component useful for realizing this collaborative paradigm is the understanding of human and robot 3D poses using non-invasive systems. Therefore, in this paper, we propose a novel vision-based system leveraging depth data to accurately establish the 3D locations of skeleton joints. Specifically, we introduce the concept of Pose Nowcasting, denoting the capability of the proposed system to enhance its current pose estimation accuracy by jointly learning to forecast future poses. The experimental evaluation is conducted on two different datasets, providing accurate and real-time performance and confirming the validity of the proposed method on both the robotic and human scenarios.

Adversarial attacks pose a significant threat to the reliability of biometric systems, particularly in security-critical applications such as identity verification and access control. Ensuring robustness against such attacks is essential for the safe deployment of face recognition technologies in real-world scenarios. To advance this goal, the 2025 Adversarial Attack Challenge for Secure Face Recognition was organized as part of the International Joint Conference on Biometrics (IJCB) 2025.The competition focused on two main tracks: Detection, where the objective was to determine whether a given face image is clean or adversarial, and Resilience, which aimed to evaluate recognition systems under adversarial perturbations. Participants were provided with a standardized dataset derived from CelebA and LFW, encompassing both clean samples and adversarial images crafted using ten diverse attack methods targeting evasion and impersonation scenarios. To ensure fairness and reproducibility, all models were trained solely on the data provided, with support from a custom open source adversarial attack package tailored for face recognition.In addition to benchmarking adversarial robustness, the challenge contributes to the research community by releasing the data set and the extensible attack package, allowing further investigation of secure and reliable face recognition systems.

Humanoid robots are increasingly being integrated into diverse scenarios, such as healthcare facilities, social settings, and workplaces. As the need for intuitive control by non-expert users grows, many studies have explored the use of Artificial Intelligence to enable communication and control. However, these approaches are often tailored to specific robots due to the absence of standardized conventions and notation. This study addresses the challenges posed by these inconsistencies and investigates their impact on the ability of Large Language Models (LLMs) to generate accurate 3D robot poses, even when detailed robot specifications are provided as input.