Comparative study for human activity recognition techniques /
Amira Ali Bebars Ali
Comparative study for human activity recognition techniques / دراسة مقارنة لطرق التعرف على النشاط البشرى Amira Ali Bebars Ali ; Supervised Elsayed E. Hemayed - Cairo : Amira Ali Bebars Ali , 2014 - 73 P. : facsimiles , photographs ; 30cm
Thesis (M.Sc.) - Cairo University - Faculty of Engineering - Department of Computer Engineering
Human activity recognition is an active area of research in computer vision with wide scale applications in video surveillance, motion analysis, and virtual reality interfaces, robot navigation and recognition, sports video analysis etc. It consists of analyzing the characteristic features of various human actions and classifying them. Part - based approach is the main focus of this thesis, a general human action recognition framework that includes spatio - temporal interest point detection, building the descriptor, constructing the codebook, and testing on the pre - trained classifier. We focus on; detectors for accurately detecting the humman action, descriptors to describe information around interest points, and classifiers for performing accurate classification
Human activity recognition MOSIFT descriptor MOSIFT detector
Comparative study for human activity recognition techniques / دراسة مقارنة لطرق التعرف على النشاط البشرى Amira Ali Bebars Ali ; Supervised Elsayed E. Hemayed - Cairo : Amira Ali Bebars Ali , 2014 - 73 P. : facsimiles , photographs ; 30cm
Thesis (M.Sc.) - Cairo University - Faculty of Engineering - Department of Computer Engineering
Human activity recognition is an active area of research in computer vision with wide scale applications in video surveillance, motion analysis, and virtual reality interfaces, robot navigation and recognition, sports video analysis etc. It consists of analyzing the characteristic features of various human actions and classifying them. Part - based approach is the main focus of this thesis, a general human action recognition framework that includes spatio - temporal interest point detection, building the descriptor, constructing the codebook, and testing on the pre - trained classifier. We focus on; detectors for accurately detecting the humman action, descriptors to describe information around interest points, and classifiers for performing accurate classification
Human activity recognition MOSIFT descriptor MOSIFT detector