View article

Recognizing the activities of humans through computer vision techniques is an important area of research. This area of research leads to various applications such as patient monitoring, fall detection, surveillance and human-computer interface. The capability for recognizing these acts lays foundation for developing highly intelligent and decision making systems. Generally, most of the mentioned applications requires automatic recognition of high-level activities, consisting of simple actions of multiple persons. Usually, the intelligence to the system is delivered only if these activities are properly classified. This paper addresses various machine learning algorithms used in classifying various activities such as Multi-Layer Perceptron, Random Forest, Naïve Bayes and SVM algorithms. This paper provides classification of general to complex human activities through comparison study and performance evaluation of these mentioned algorithms using very large set of images. This review will provide much needed information for further research in more productive areas.