Using 3D Convolutional Neural Network in Surveillance Videos for Recognizing Human Actions

Human action recognition is a very important component of visual surveillance systems. The demand for automatic surveillance systems play a crucial role in the circumstances where continuous patrolling by human guards are not possible. The analysis in surveillance scenarios often requires the detection of certain specific human actions. The automated recognition of human actions in detecting certain human actions are considered here. The main aim is to develop a novel 3D Convolutional Neural Network (CNN) model for human action recognition in realistic environment. The features are extracted from both the spatial and the temporal dimensions by performing 3D convolutions, by which, capturing the motion information encoded in multiple adjacent frames. The evolved model generates multiple information from the input frames, and the information from all the channels are combined and that is to be the final feature. The developed model automatically tends to recognize specific human actions which needs attention in the real world environment like in pathways or in corridors of any organization. This proposed work is well suitable for the situations like where continuous patrolling of humans are not possible, to prevent certain human actions which are not allowed inside the organisation premises.