Violent Behavioral Activity Classification Using Artificial Neural Network

Abstract

Detection of video information is a great help in classifying non-standard / abnormal human behavior. It is more difficult to detect objects from videos when information in videos are time bound to each other. In this paper we discuss the need to detect and classify this data. Also, we try to improve classification process by various methods. A specially modified convolution neural network architecture was used along with Long Short-Term Memory (LSTM) and time distribution in experiment. Convolution neural layers for 2D data, in architecture were used.