An Approach of Understanding Human Activity Recognition and Detection for Video Surveillance using HOG Descriptor and SVM Classifier
Chandrashekar M Patil, Jagadeesh Basavaiah, M. Meghana · 2017 International Conference on Current Trends in Computer, Electrical, Electronics and Communication (CTCEEC) · 2017
In this video surveillance moving object detection and recognition is the important research area of computer vision. Detection and recognition of moving is not easy task as continuous deformation of objects takes place during movement. Any moving objects has several attributes in temporal and spatial spaces. In spatial space object vary in size where as in temporal space it vary in moving speed. This work mainly focuses on multiple human detection and activity recognition. Multiple human video datasets are considered and in order to detect and track multiple human. Background subtraction technique is used for detecting moving multiple humans. Histogram of Oriented Gradient feature descriptor is used to extract features. For human activity recognition Support Vector Machine classifier is used.