DEARESt: Deep Convolutional Aberrant Behavior Detection in Real-world Scenarios

In this paper, we present a new technique: DEARESt for “Aberrant Behavior Detection in surveillance videos DEARESt employs a two-stream network to extract appearance and motion flow features separately, from a video stream. These features are concatenated to form a single feature vector that is further used to classify a video. Appearance features are captured by using VGG-19, while optical flows between successive frames are calculated and fed to FlowNet in order to extract motion features. After concatenation of features Neural Network is used for classification. Performance of proposed model is evaluated against a subset of UCF-crime dataset. From the experimental results it is evident that DEARESt outperforms state-of-art methods namely: VGG-16, VGG-19 and FlowNet.