Hybrid Classification Method for the Human Activity Detection

In person-to person collaboration and social relations, the identification of human activity plays a significant role to obtain information of the individuality of people, their character and emotional state. It is difficult to extract this information. The research of the scientific fields of computer vision and Machine Learning is useful to recognize the actions of other people. A hybrid technique is proposed in this research for the human activity recognition. This hybrid technique has integrated the Principal Component Analysis, K-mean and voting classification. The voting classification improves results in comparison with the existing logistic regression classification with regard to precision and recall etc.