Prediction of Impulsive Aggression Based on Video Images.

Area of Science:

• Forensic Psychology
• Biomedical Engineering
• Computer Vision

Background:

• Existing attack behavior prediction methods suffer from subjectivity, low accuracy, and high concealment.
• There is a need for objective and accurate methods for impulsive aggression prediction.

Purpose of the Study:

• To propose and validate a video-based impulsive aggression prediction method integrating physiological parameters and facial expression information.
• To address the limitations of current prediction techniques.

Main Methods:

• Utilized imaging equipment to capture video and facial expression data.
• Employed imaging photoplethysmography (IPPG) to obtain heart rate variability parameters.
• Developed a ResNet-34 model for facial expression recognition and a random forest model for prediction.

Main Results:

• The proposed method achieved an accuracy of 89.39% in predicting impulsive aggression.
• Demonstrated the feasibility of using physiological and facial expression data for prediction.
• Validated through an impulsive aggression induction experiment.

Conclusions:

• The integrated approach of physiological parameters and facial expressions offers a feasible and accurate method for impulsive aggression prediction.
• This research holds significant theoretical and practical value for developing new prediction methods.
• The method has potential applications in safety monitoring in high-risk environments like prisons and rehabilitation centers.