Human Activity Recognition Method for Moderate Violence Detection

2026-05-04 · Source: Computer Vision and Pattern Recognition · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Computer Vision · Depth: Advanced, quick

Summary

This research introduces an automated system designed for real-time detection of moderate physical violence, specifically pushing, using surveillance camera footage. The system integrates YOLO11 for human detection and YOLO11-Pose for extracting skeletal keypoints. It calculates body inclination and joint angles between shoulders and hips, feeding this data into a Random Forest classifier to differentiate between normal behavior and aggressive physical contact. Evaluated across three case studies, the model achieved a precision of 0.98 in controlled, frontal-view environments. Even in challenging real-world scenarios with high-altitude, steep-angle recordings and significant perspective distortion, the system maintained a precision of 0.72, demonstrating the viability of skeletal analysis for early violence intervention in urban security.

Key takeaway

For urban security professionals and computer vision engineers developing public safety solutions, this research indicates that real-time moderate violence detection is feasible using skeletal analysis. You should consider integrating YOLO11/YOLO11-Pose with Random Forest classifiers for robust performance, even under challenging surveillance conditions, to enable proactive intervention and enhance public safety.

Key insights

Skeletal keypoint analysis with Random Forest can detect moderate violence in real-time surveillance.

Principles

• Early detection prevents escalation.
• Skeletal analysis identifies aggressive contact.

Method

The method uses YOLO11/YOLO11-Pose for detection and keypoint extraction, then a Random Forest classifier on body inclination and joint angles to classify behavior.

In practice

• Deploy in public spaces for early intervention.
• Integrate with existing surveillance systems.

Topics

• Human Activity Recognition
• Moderate Violence Detection
• YOLO11
• Skeletal Keypoint Extraction
• Random Forest Classifier

Best for: AI Scientist, Computer Vision Engineer, Research Scientist

Related on AIssential

• Interpretable Human Activity Recognition for Subtle Robbery Detection in Surveillance Videos — A pose-driven hybrid system detects subtle snatch-and-run robberies in real-time using kinematic and interaction features.
• YOLO-AMC: An Improved YOLO Architecture with Attention Mechanisms for Building Crack Detection — Integrating attention mechanisms into YOLO's feature fusion significantly improves crack detection accuracy and deployment efficiency.
• Small Model, Big Results: How a 5MB On-Device NSFW Detector Outperforms Cloud APIs and Passes a… — Small, on-device object detection models can surpass large cloud APIs in NSFW accuracy and fairness by focusing on anatomical shapes.
• Unsupervised Skeleton-Based Action Segmentation via Hierarchical Spatiotemporal Vector Quantization — A hierarchical spatiotemporal vector quantization method improves unsupervised skeleton-based action segmentation.
• Comparative Analysis of Military Detection Using Drone Imagery Across Multiple Visual Spectrums — Diverse visual spectrum datasets significantly enhance drone-based military object detection robustness in varied environments.
• A Machine Learning Framework for Real-Time Personalized Ergonomic Pose Analysis — Real-time personalized ergonomic pose analysis leverages 3D volumetric data to overcome 2D camera limitations and improve workplace safety.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Computer Vision and Pattern Recognition.