Deep Multi-Scale Features Fusion for Effective Violence Detection and Control Charts Visualization.

Area of Science:

• Computer Vision
• Artificial Intelligence
• Process Control

Background:

• Automated video surveillance systems are crucial in real-world CCTV environments, with a primary focus on enhancing accuracy.
• Current systems require more attention towards integrating surveillance expert assistance for effective data analysis and rapid decision-making using advanced computer vision algorithms.

Purpose of the Study:

• To introduce control charts, a process control technique, for the analysis of surveillance video data.
• To develop a novel deep learning-based violence detection framework that merges control charts with advanced algorithms.
• To enhance the accuracy and effectiveness of automated surveillance systems by incorporating spatial and temporal information fusion.

Main Methods:

• A novel deep learning framework was developed for violence detection, integrating control charts for data analysis.
• The framework uniquely considers both spatial and temporal representations of video data.
• A multi-scale strategy was employed to fuse spatial information with the temporal dimension of the deep learning model at multiple levels.

Main Results:

• The proposed technique successfully integrates spatial and temporal information for robust violence detection.
• Control charts were utilized to maintain a history of surveillance video analysis results, validating risk levels.
• Experimental results on existing datasets and real-world data confirm the approach's prominence in automated surveillance.

Conclusions:

• The study presents a novel approach combining control charts and deep learning for automated video surveillance and violence detection.
• The proposed method offers a significant advancement in analyzing surveillance data, providing pre- and post-analyses of violent events.
• This research highlights the potential of process control techniques in improving the efficiency and reliability of intelligent surveillance systems.