Detection of Anomalous Behavior in Prisoner Gatherings Using Surveillance Cameras
Research presentation and project archive
Archived academic material · Original source · Resource library
This archived description records a research concept or project as presented in the source. Proposed outcomes and implementation status have not been independently verified.
With Emphasis on Artificial Intelligence Applications
- Hadi Sadoghi Yazdi : PhD in Electronics, Expert Consultant in Machine Vision/Learning and Data Systems
- Affilation and institute:
- Professor of Electrical and Computer Engineering, Ferdowsi University of Mashhad
- Director of Pattern Recognition Laboratory
- Member of SCIIP - Center of Excellence on Soft Computing and Intelligent Information Processing
<picture><source type="image/webp" srcset="/assets/img/legacy/prison-ai-security-system-nature-inspired-55884ecba5-480.webp 480w, /assets/img/legacy/prison-ai-security-system-nature-inspired-55884ecba5-800.webp 800w, /assets/img/legacy/prison-ai-security-system-nature-inspired-55884ecba5-1400.webp 1400w" sizes="(min-width: 930px) 900px, 95vw"><img loading="lazy" src="/assets/img/legacy/prison-ai-security-system-nature-inspired-55884ecba5.jpg" alt="Identification of Behavior in Prison with AI"></picture>
Identification of Behavior in Prison with AI
Table of Contents
- With Emphasis on Artificial Intelligence Applications
- Table of Contents
- Introduction
- Methodology
- Challenges and Mitigations
- Expected Outcomes
- Additional Service Offerings
- References
- Contact Me
Introduction
Objectives
- Develop AI-powered system for real-time anomaly detection in prison CCTV feeds.
- Reduce response times to threats (e.g., fights, self-harm).
- Ensure ethical compliance (privacy, bias mitigation).
- Achieve 90%+ accuracy in detecting crowd-based anomalies.
<picture><source type="image/webp" srcset="/assets/img/legacy/objective-5a25733afc-480.webp 480w, /assets/img/legacy/objective-5a25733afc-800.webp 800w, /assets/img/legacy/objective-5a25733afc-1400.webp 1400w" sizes="(min-width: 930px) 900px, 95vw"><img loading="lazy" src="/assets/img/legacy/objective-5a25733afc.jpg" alt="AI-based real-time anomaly detection in prison CCTV"></picture>
AI-based real-time anomaly detection in prison CCTV
Existing Implementations in Various Countries
- China: AI surveillance with facial recognition for threat detection (prison breaks impossible).
<picture><source type="image/webp" srcset="/assets/img/legacy/china-cf1c87ce37-480.webp 480w, /assets/img/legacy/china-cf1c87ce37-800.webp 800w, /assets/img/legacy/china-cf1c87ce37-1400.webp 1400w" sizes="(min-width: 930px) 900px, 95vw"><img loading="lazy" src="/assets/img/legacy/china-cf1c87ce37.png" alt="Stephen Chenin Beijing, 1 Apr 2019"></picture>
Stephen Chenin Beijing, 1 Apr 2019
- UK: Avigilon (software for contraband and behavior monitoring). Avigilon’s self-learning AI continuously improves detection accuracy by adapting to environments, reducing false alarms and enhancing security threat identification.
- Singapore: AI-based CCTV for fights and headcount checks.
- India: Facial recognition in Tihar jail for anomaly detection.
Singapore: AI-based CCTV for fights and headcount checks. New technology on trial at Changi Prison can detect cell fights through video analytics.
<picture><source type="image/webp" srcset="/assets/img/legacy/changi-singapour-0e52744ff8-480.webp 480w, /assets/img/legacy/changi-singapour-0e52744ff8-800.webp 800w, /assets/img/legacy/changi-singapour-0e52744ff8-1400.webp 1400w" sizes="(min-width: 930px) 900px, 95vw"><img loading="lazy" src="/assets/img/legacy/changi-singapour-0e52744ff8.png" alt="Fights and headcount checks"></picture>
Fights and headcount checks
India: Tihar Jail, one of the largest prison complexes in India, has been implementing a facial recognition, anomaly system and enhance security.
<picture><source type="image/webp" srcset="/assets/img/legacy/jail-486d1251dd-480.webp 480w, /assets/img/legacy/jail-486d1251dd-800.webp 800w, /assets/img/legacy/jail-486d1251dd-1400.webp 1400w" sizes="(min-width: 930px) 900px, 95vw"><img loading="lazy" src="/assets/img/legacy/jail-486d1251dd.png" alt="Tihar installs 1,248 CCTVs with facial recognition"></picture>
Tihar installs 1,248 CCTVs with facial recognition
Methodology
Data Collection
- Gather anonymized CCTV footage (ethical approvals).
- Label data for normal vs. anomalous behaviors (e.g., fights, gatherings).
- Example: Deep Learning for Video Anomaly Detection: A Review
<picture><source type="image/webp" srcset="/assets/img/legacy/datacollectionreview-ea78be5318-480.webp 480w, /assets/img/legacy/datacollectionreview-ea78be5318-800.webp 800w, /assets/img/legacy/datacollectionreview-ea78be5318-1400.webp 1400w" sizes="(min-width: 930px) 900px, 95vw"><img loading="lazy" src="/assets/img/legacy/datacollectionreview-ea78be5318.jpg" alt="Illustrations of frame-level (Top) and pixel-level (Bottom) output"></picture>
Illustrations of frame-level (Top) and pixel-level (Bottom) output
Model Development
- Frame Analysis
- Use CNNs for video frame analysis, RNNs for temporal tracking.
- Frame Encoding
- EfficientNet-B7 (pretrained on Kinetics-700) for high-resolution feature extraction.
- 3D Convolutional Blocks (I3D) for short-term spatiotemporal features (5-frame snippets).
-
Temporal Context
- Bidirectional Quasi-Recurrent Neural Networks (QRNNs): Lightweight alternative to LSTMs/GRUs. Processes sequences with parallel convolution + recurrent pooling (reducing latency 40% vs. GRU).
- Attention Mechanisms: Self-attention layers to weight critical frames (e.g., sudden motion/occlusion).
- Implement anomaly detection
- use YOLO v7, Autoencoders
- use Adaptive frame skipping based on optical flow magnitude.
- Edge Computing Deployment
- Hardware: NVIDIA Jetson AGX Orin (48 TOPS) / Google Coral TPU.
- Model Distillation: Teacher (EfficientNet-B7 + QRNN) → Student (MobileNetV3 + QRNN).
<picture><source type="image/webp" srcset="/assets/img/legacy/deepnet-237819c6ab-480.webp 480w, /assets/img/legacy/deepnet-237819c6ab-800.webp 800w, /assets/img/legacy/deepnet-237819c6ab-1400.webp 1400w" sizes="(min-width: 930px) 900px, 95vw"><img loading="lazy" src="/assets/img/legacy/deepnet-237819c6ab.png" alt="Deep Neural Network"></picture>
Deep Neural Network
Deployment
- Deploy on CCTV with AI accelerators (e.g., NVIDIA Jetson).
- Pilot testing in controlled prison settings.
- Conduct ethical audits (GDPR (General Data Protection Regulation)-like policies, bias checks).
<picture><source type="image/webp" srcset="/assets/img/legacy/deployment1-38b70f9cb0-480.webp 480w, /assets/img/legacy/deployment1-38b70f9cb0-800.webp 800w, /assets/img/legacy/deployment1-38b70f9cb0-1400.webp 1400w" sizes="(min-width: 930px) 900px, 95vw"><img loading="lazy" src="/assets/img/legacy/deployment1-38b70f9cb0.png" alt="Deployment of AI-enabled CCTV system on-site with automotive-grade accelerators for real-time monitoring."></picture>
Deployment of AI-enabled CCTV system on-site with automotive-grade accelerators for real-time monitoring.
Technology Stack
| Component | Description |
|---|---|
| Hardware | High-res CCTV with IR, motion sensors |
| Software | Python (OpenCV, TensorFlow), MQTT alerts |
| AI Models | CNN-RNN, YOLO, GANs for synthetic data |
| Security | Encrypted storage, access controls |
<picture><source type="image/webp" srcset="/assets/img/legacy/technologystackvisualization-1f8c02eaf6-480.webp 480w, /assets/img/legacy/technologystackvisualization-1f8c02eaf6-800.webp 800w, /assets/img/legacy/technologystackvisualization-1f8c02eaf6-1400.webp 1400w" sizes="(min-width: 930px) 900px, 95vw"><img loading="lazy" src="/assets/img/legacy/technologystackvisualization-1f8c02eaf6.png" alt="Technology Stack Visualization"></picture>
Technology Stack Visualization
Challenges and Mitigations
- Privacy Concerns: GDPR-like policies, data anonymization.
- AI Bias: Diverse training data, fairness algorithms.
- False Positives: Fine-tune models with prison-specific data.
<picture><source type="image/webp" srcset="/assets/img/legacy/concerns-23f4a27d8c-480.webp 480w, /assets/img/legacy/concerns-23f4a27d8c-800.webp 800w, /assets/img/legacy/concerns-23f4a27d8c-1400.webp 1400w" sizes="(min-width: 930px) 900px, 95vw"><img loading="lazy" src="/assets/img/legacy/concerns-23f4a27d8c.jpg" alt="Privacy, fairness, and accuracy: GDPR-like privacy with anonymization, bias mitigation through diverse data and fairness algorithms, and reduced false positives via prison-specific fine-tuning."></picture>
Privacy, fairness, and accuracy: GDPR-like privacy with anonymization, bias mitigation through diverse data and fairness algorithms, and reduced false positives via prison-specific fine-tuning.
Expected Outcomes
- 30-50% reduction in security incidents (inspired by India).
- Cost savings via automated monitoring (like Singapore).
- Scalable system for global adoption.
<picture><source type="image/webp" srcset="/assets/img/legacy/expectedoutcomes-be381cfd96-480.webp 480w, /assets/img/legacy/expectedoutcomes-be381cfd96-800.webp 800w, /assets/img/legacy/expectedoutcomes-be381cfd96-1400.webp 1400w" sizes="(min-width: 930px) 900px, 95vw"><img loading="lazy" src="/assets/img/legacy/expectedoutcomes-be381cfd96.png" alt="Significant security gains, cost savings, and scalable global deployment"></picture>
Significant security gains, cost savings, and scalable global deployment
Additional Service Offerings
- Smart Prisons: Closing the gap between smart-city innovations and their application in correctional facilities.
- Tackle Prison Contraband: Trial of new technology to tackle prison contraband.
- Guarding the Perimeter: Drone Detection and Neutralization in Prisons
<picture><source type="image/webp" srcset="/assets/img/legacy/drone-3b05c5d8d7-480.webp 480w, /assets/img/legacy/drone-3b05c5d8d7-800.webp 800w, /assets/img/legacy/drone-3b05c5d8d7-1400.webp 1400w" sizes="(min-width: 930px) 900px, 95vw"><img loading="lazy" src="/assets/img/legacy/drone-3b05c5d8d7.jpg" alt="Protect Your Correctional Facility From Drone Threats"></picture>
Protect Your Correctional Facility From Drone Threats
References
- AI surveillance systems in China, Hong Kong, Singapore (2023 reports).
- UK Altcourse prison: Avigilon software (2022).
- India Punjab jails: AI CCTV deployment (2024).
Contact Me
<p><strong>Hadi Sadoghi Yazdi</strong></p>
<p>AI and Data Specialist</p>
<p>Email: <a href="mailto:h-sadoghi@um.ac.ir">h-sadoghi@um.ac.ir</a></p>
<p>Phone: <a href="tel:+985138805117">+98-51-38805117</a></p>
<p>
Website:
<a href="https://h-sadoghi.github.io/" target="_blank">https://h-sadoghi.github.io/</a> |
<a href="/" target="_blank">https://hadisadoghiyazdi1971.github.io/</a>
</p>