louvre-object-detection

ART WATCHER – Real-Time Artifact Security System

🛡️ Overview

ART WATCHER is an intelligent, multi-laptop security solution designed for museums, galleries, and high-value asset facilities. It combines AI-powered object detection, automated movement alerts, audio-triggered recordings, and live monitoring to prevent theft in real-time.

---

🚀 Key Features

1. Real-Time Artifact Tracking

• YOLOv8 neural network for object detection, optimized with OpenVINO.
• Detects and tracks multiple high-value objects simultaneously.
• Configurable movement thresholds for precise alerts.
• Instant notifications when objects are moved or removed.

2. Multi-Laptop Architecture

• Detection Laptop: Runs the AI detection engine and triggers alerts.
• Audio-Triggered Recorder Laptop: Listens for alert sounds and records video clips automatically.
• Web Dashboard Laptop: Displays live object status, movement alerts, and timestamps.

3. Automated Evidence Capture

• Video and images captured automatically on theft or movement detection.
• Stored securely with timestamps.
• Preserves chain-of-custody for legal proceedings.

4. Live Web Monitoring

• Accessible dashboard showing object status, last seen time, movement history, and connection health.
• Sub-second updates (500ms) for near real-time monitoring.
• Color-coded alerts for immediate situational awareness.

---

🏗️ System Architecture

┌───────────────┐      ┌──────────────────┐      ┌─────────────────┐
│ Detection Cam │─────▶│ Detection Laptop │─────▶│ Web Dashboard   │
│  (Artifacts)  │      │ (YOLOv8 + OV)    │      │ (Flask + JSON)  │
└───────────────┘      └──────────────────┘      └─────────────────┘
         │
         ▼
┌───────────────────────┐
│ Audio-Triggered Laptop│
│ (Records Video Clips) │
└───────────────────────┘

---

🛠️ Technical Stack

• Computer Vision: OpenCV, YOLOv8, OpenVINO
• Backend: Python, Flask
• Audio Alerts: Pygame
• Frontend: HTML5, CSS3, JavaScript
• Interprocess Communication: JSON files

---

📦 Installation & Setup

Prerequisites

# Python 3.8+
python --version

Setup

# Clone the repo
git clone <repo-url>
cd art-watcher

# Install dependencies
pip install opencv-python numpy pygame ultralytics openvino flask

# Add alert sound
# Place "AGAIN_fetty.mp3" in the project directory

# Test camera
python -c "import cv2; cap = cv2.VideoCapture(0); print('Camera OK' if cap.isOpened() else 'Camera Error')"

---

🎮 Usage

1. Start Detection Laptop

python detection.py

• Launches YOLOv8 object detection and real-time movement alerts.
• Tracks multiple bottles and plays alert sound on movement.

2. Start Web Dashboard Laptop

python web_server.py

• Access dashboard at http://localhost:5000
• Displays object presence, last seen, last movement, and connection status.

3. Start Audio-Triggered Recorder Laptop

• Listens for alert sound and records clips automatically for evidence.

---

🔒 Security Features

• Local network operation – no internet dependency.
• Encrypted and timestamped evidence storage.
• Redundant multi-camera monitoring.
• Instant alerts with <100ms response time.
• Audio-triggered recording ensures evidence even if direct detection is missed.

---

📈 Future Enhancements

• Multi-artifact tracking per display case.
• Facial recognition for suspect identification.
• Suspicious behavior prediction.
• Mobile push notifications for alerts.
• Cloud backup of evidence.
• Integration with building access control and thermal imaging.

---

👀 Demo Flow

1. Display bottle → Dashboard shows Object Present ✅
2. Move bottle → Audio alert triggers + dashboard shows Movement Detected 🚨
3. Remove bottle → Dashboard shows OBJECT MISSING ⚠️
4. Secondary laptop records video clip automatically.
5. Evidence stored securely with timestamps.

---

"The best heist is the one that never happens." – ART WATCHER Team