Violence detection using affective features from upper body expressions

• Main Authors: Baloch, Saba, Syed Abu Bakar, Syed Abdul Rahman, Mohd. Mokji, Musa, Waseem, Saima
• Format: Conference or Workshop Item
• Published: 2023
• Subjects: TK Electrical engineering. Electronics Nuclear engineering
• Online Access: http://eprints.utm.my/107869/; http://dx.doi.org/10.1145/3604078.3604116

In recent years, places such as elderly care houses, nursing homes, schools, prisons, psychic wards are prone to aggressive and violent and abusive interactions. This leads to the need for a system that could automatically detect such abhorrent behaviours. With the rapid increase in the use of surveillance cameras, the implementation of such a system can help in deterring such intolerable incidents and thus providing a much safer environment. In the paper, a violence detection system is proposed using affective features for the anger emotion in order to enhance to efficiency of the existing violence detection systems. OpenPose toolbox is used for extracting 2D skeleton keypoints from the RWF2000 dataset and MATLAB Classification Learners app is used for classification, producing encouraging results.