期刊名称：International Journal of Innovative Research in Science, Engineering and Technology
印刷版ISSN：2347-6710
电子版ISSN：2319-8753
出版年度：2016
卷号：5
期号：10
页码：17713
DOI：10.15680/IJIRSET.2016.0510032
出版社：S&S Publications
摘要：Detecting human action is a part of computer vision is a challenging problem. Many machine learningmethodologies have been adapted to implement this problem. Most of the research has been towards human actionrecognition via moving images or video frames. Detecting human action finds applications in robotic vision,surveillance system etc.The main concept behind action detection is to find out the body shape, pose, angle etc. features. Once these featuresare obtained any machine learning algorithm can be used to classify the objects in action. But this problem seems to bedifficult in still images where there are challenges posed by background of the image. In still images there are nodisplacements in the body pose. Hence finding action in still images is a challenge. In this thesis work, I propose amethod of still images based action recognition system through support vector machines. Support vector machines arelinear classifier. Thus it is again a challenge to modify the model of linear svm and train it to classify multiple classes’data set.
