電子情報通信学会総合大会講演要旨
D-12-74
Hand Detection Using One-Class SVM
○Harry Haryadi・Keisuke Kameyama(The Univ. of Tsukuba)
Detecting hand from still images has been a challenging task in the computer vision field due to its shape variability when observed from different viewpoints. However hand position in an image holds important information for human layout analysis which can benefits behavioral, action analysis and interpretation of sign language. In sequential image a hand position can be estimated based on its movement and historical data using a tracking algorithm, on the opposite on still image, detection become more challenging due to the lack of supporting data. In this work, we propose hand detector built from one-class SVM classifier and Histogram of Oriented Gradient as feature descriptor.