Violence Detection Using Spatiotemporal Features with 3D Convolutional Neural Network.

Area of Science:

• Computer Science
• Artificial Intelligence
• Security Systems

Background:

• Increasing use of surveillance cameras in smart cities generates vast data for monitoring.
• Automatic detection of violent activities is crucial for preventing social, economic, and ecological damage.

Purpose of the Study:

• To propose a triple-staged, end-to-end deep learning framework for automatic violence detection.
• To enhance security systems in smart cities, schools, and hospitals through rapid identification of abnormal events.

Main Methods:

• Utilized a lightweight convolutional neural network (CNN) for initial person detection.
• Employed a 3D CNN to extract spatiotemporal features from video sequences.
• Optimized the 3D CNN model using Intel's OpenVINO toolkit for efficient execution.

Main Results:

• The proposed framework achieved superior performance compared to existing state-of-the-art methods on benchmark datasets.
• The system successfully detects violent activities and transmits alerts for prompt action.

Conclusions:

• The developed deep learning framework offers an effective solution for automatic violence detection.
• This technology can significantly assist security departments in maintaining public safety and preventing harm.