PREDICTION OF UNUSUAL ACTIVITY USING DEEP LEARNING

In this paper effective methods for identifying abnormalities in the videos are compared with the different methods. Currently uses of convolutional neural networks has exposed potentials of convolutional network layers for object recognition and detection especially in images. However, convolutional neural networks are controlled and wanted layers as signs of learning. In RNN architecture, includes two key components, one for learning the temporal evolution of the spatial features and other for spatial feature representation. In this context the comparison of the different types of techniques is made with the recurrent neural networks. RNN gives the better performance when compared to other Methods.