Annotation Contribution to Classification Accuracy of Person Identification Based Gait Biometric

Abstract

Annotation takes part in person identification based gait. Many studies use annotation in gait analysis for person identification using silhouette technique. However, implementation of annotation in 3D gait data such motion capture is still rare, but it is comprimising as study for person recovery which utilizes only certain human body parts. To begin this study, a person identification using classification technique is used as study case. Annotation which consists of binary decision making (BD) and rectangular rounded (RR) are performed to limit body part area that is selected to be processed by Naïve Bayessian classification. No annotation is also utilized as comparation those two annotation techniques. By using 6, 10, and 16 markers usage scenarios, result shows that BD is always outperform to no annotation, while RR has lower accuracy to no annotation at using 10 markers. Accuration gap analysis shows that comparation between BD and RR shows no consistency rate on each amount of markers usage.