Biometric recognition and presentation attack detection (PAD) methods strongly rely on deep learning algorithms. Though often more accurate, these models operate as complex black boxes. Interpretability tools are being used to delve deeper into the operation of these methods and we advocate their integration in the PAD scenario.The work presented will focus on face PAD CNN-based models analysis through evaluation with traditional PAD metrics and with interpretability tools. We will go through the topic and recent works that confirm the urge to establish new approaches in biometrics that incorporate interpretability tools
Curriculum vitæ
Ana F. Sequeira is a researcher focused on Computer Vision and Machine Intelligence for BIOMETRICS. Currently, at INESC TEC, Porto, Portugal; previously at IrisGuard UK Ltd, UK; and at University of Reading, UK. Holds a PhD in Electrical and Computing Engineering; a MSc and a Degree in Mathematics, all from the University of Porto. Ana’s research comprises xAI for biometrics; liveness detection; biometric recognition for border control; as well as facial analysis topics, such as emotion recognition; among others.Ana led the construction of biometric databases; managed biometric competitions and co-authored research publications.
We thank our media partner: