A Proficient Extreme Learning Machine Approach For Tracking And Estimating Human Poses.

Abstract

Tracking and Estimation of human pose in real time is a difficult problem with many interesting applications. Automated tracking is useful in variety of domains including human computer interaction, gait analysis, the film industry and entertainment. The existing system uses different algorithm to estimate and track human poses but the limitation is due to the error rate which is above 10%. In the proposed system effective filtering technique and background subtraction technique is used in order to remove clutters and noises. The objects in each frame are tracked and the corner points are identified. The corner points are used to identify objects that are human. Finally, Extreme Learning Machine classifier is used to identify and estimate the exact human pose from video