An FPGA-based IP for recognizing violence against children.

We present in this paper an FPGA-based IP for recognizing most common violent actions against children (VACR IP). VACR IP uses only skeleton joints data as inputs to keep the privacy of people inside their homes or in the schools. The proposed hardware achieved a detection rate of 97.72%, processing speed 761FPS, and a latency value equals to 2.79msec using a 50MHz system clock. In sum, this research method presents:•First FPGA-based IP for recognizing most common child abuses without any privacy breach of people real images by using only skeleton joints data.•The IP can detect the violence and the type of the violent action.•The IP can be embedded in complete systems to be installed in schools by school psychologists and counselors.

Vision-based system model for detecting violence against children.

We present in this paper a machine learning model for detecting violence against children. This model, which uses skeletal data acquired by depth sensors achieved a high accuracy violence detection rate of 99.03 %. In sum, this research method presents: •First ML-based method for detecting most common child abuses, which keeps the privacy of people by using only skeleton joints data.•The model has only two classes (violent action, non-violent action).•The model can be a base for other researches and implementations in schools by school psychologists and counselors.

Violent actions against children.

We present in this paper a novel dataset (MMU VAAC) for violent actions against children recognition. This original dataset has been recorded using Microsoft Kinect with the usage of a child mannequin. MMU VAAC dataset contains skeleton joints, depth, and RGB modalities.