Identifying Criminal Activites in Video Footages using Deep Learning Methods
Abstract
Identifying Criminal Activities from video footages is the process of getting high level features from a set of training data and analyzing it to understand the related features which contribute to a video being labelled as “Criminal” or “Normal”. We used image processing techniques to extract features from the related videos and the machine was able to learn the features to able to correctly classify the given video as “Criminal” or “Normal”. The proposed system uses the following tools – a network of Convolutional Neural Network layers amalgamated with Artificial neural network which uses backpropagation to converge to a minimum loss.