Unconstrained data acquisition frameworks and protocols


The identification of humans in non-ideal conditions has been gaining increasing attention, mainly supported by the multitude of novel methods specifically developed to address the covariates of non-cooperative biometric recognition. Surveillance environments are one of the most representative cases of unconstrained scenarios, where fully automated human recognition has not been achieved, yet. These environments are particularly harsh for several reasons (e.g., high variations in illumination, pose, and expression), but the limited resolution of the biometric data acquired is regarded as the major factor for performance degradation. Consequently, different strategies for imaging subjects at-a-distance have been introduced. This chapter provides a comprehensive review of the state-of-the-art surveillance systems for acquiring biometric data at-a-distance in a non-cooperative manner. The challenges and the open issues of current architectures are highlighted. Also, the most promising strategies and future lines of research are outlined.

Human Recognition in Unconstrained Environments Using Computer Vision, Pattern Recognition and Machine Learning Methods for Biometrics 2017, Pages 1-30
Silvio Barra
Silvio Barra
Assistant Professor

Assistant Professor @ University of Naples