Biometrics has been a thriving field of Pattern Recognition forlong. Both Academia and Business have thus focused their at-tention in the practical use of biometrics to advance and pro-mote varying applications in forensics, security and surveillance,health-care, mobility, human- computer interaction (HCI), safetyand trust, and automation and robotics all of much interest toGovernment, Finance, Education etc. As the biometric problemsthat have to be solved have become ever more challenging andsophisticated, the techniques involved have found help and inspi-ration in human perception. As such, soft biometrics are traitsthat are naturally used by humans to distinguish their peers. En-hance as well both identification and re-identification but avoidimpersonation and disinformation. Soft biometrics are physicaland behavioral traits that capture human characteristics that gobeyond appearance, e.g., age, gender, and gait. This thesis aims toadvance the state-of-the art in varying applications on how to es-timate the head pose of a subject and assist in her face recognitionincluding tributaries such as attention, gait analysis to estimatethe gender, and human behavior to meter the extent of cooperationand interest. Complete processing pipelines including data cap-ture, preprocessing and feature extraction, adaptation and clas- sification, and decision making are presented and comparativelyevaluated to show merit and advantage compared with currentstate-of-the art methods. Such methods are further integratedand successfully applied among others to the purposeful interplaybetween social engineering and social humanoid robots. [edited by Author]

XXXIII ciclo

2019 - 2020