Audio Based Violent Scene Detection Using Extreme Learning Machine Algorithm

In, this paper we proposed the work which addresses to a methodology for Audio Based Violent Scene Detection by practicing distinct features and classifier using two different datasets. Classifier Extreme Learning Machine (ELM) for Violent Scene Detection (VSD) using Temporal and Frequency domain audio features is demonstrated. As the violent scene detection in videos has realistic importance in a number of applications, such as sensible surveillance, video retrieval, Internet filtering, film rating, toddler protection towards violent behavior and so on. Hence, in digitalize form needs a Violence Detection system. For this study we use, two types of datasets Standardized Media Eval VSD-2014 and other is Customized dataset for the same ELM classifier. After successful training and testing, 85.7% accuracy is achieved by ELM for VSD-2014 dataset and 88.89% for Customized dataset respectively.