Deep NeuralNet For Violence Detection Using Motion Features From Dynamic Images

Violent Action Recognition has been a challenging topic in monitoring human activities, especially in public premises. Deep neural nets and Transfer learning have proven highly successful in the detection of violent activities. In this paper, a novel deep NeuralNet system is proposed for the task of Detecting Violence by extraction of motion features from RGB Dynamic Images (DI). Motion feature extraction and prediction of violent content using a stream of RGB DI is done effectively by fine-tuning the pre-trained Inception-Resnet-V2 model. Improved and highly effective methods for intelligent analysis are highly demanded. For performance validation of the proposed novel model, tests are performed on three popular and publically available benchmarks - Hockey Fight dataset, Real Life Violence Dataset, and movie dataset.