Abstract:
Nowadays, real-time information is very
important and learning based human motion has
fascinated range from detection to tracking state in
Computer Vision. In this system, the real-time videos
are used to detect, track, and classify object or events in
order to understand a real-world scene. Video based
real time human motion detection and tracking is a
complex and challenging task due to variation in
human pose, shape variation, illumination changes
and background appearance. A real-time mechanism
is to detect the person and their moving within an
environment from the video camera. This paper
proposes human motion detection from video
sequences. The proposed method includes three stages:
human detection, motion tracking and accuracy result
based on learning approach. The result is to become an
efficient detection system for real-time human motion.
Motion detection and tracking is determined by using
Histogram of Oriented Gradients (HOG) feature
extractor and Support Vector Machine (SVM) detector
with learning human pattern which is well performed
human detection and tracking in video sequences.
Detailed analysis is carried out on the performance
and accuracy of the system with the various test videos
to show the results. The experimental results
demonstrate the efficiency of the method
