SSF-Vision 👁️🛡️

Intelligent Surveillance System for Real-Time Threat Detection & Tracking

SSF-Vision is a comprehensive, self-hosted smart surveillance and security framework. Leveraging state-of-the-art computer vision models and machine learning techniques, it transforms standard camera feeds into an intelligent monitoring system capable of anomaly detection, automated logging, and multi-camera person re-identification (ReID).

The entire system is orchestrated through a central dashboard (main.py), making it easy to manage multiple surveillance feeds and security tools from a single interface.

---

🚀 Key Features

• Centralized Dashboard (main.py): Acts as the command center to launch and manage all other security modules (anomaly detection, tracking, etc.) from one unified interface.
• Real-Time Object & Threat Detection: Powered by YOLOv8 (yolov8n.pt), ensuring lightning-fast and highly accurate detection of people, vehicles, and objects.
• Behavioral Anomaly Detection (anomaly.py): AI-driven logic to monitor feeds for suspicious activities, unauthorized access, or unusual behavioral patterns.
• Person Re-Identification & Tracking (reid.py & suspect_finder.py): Advanced tracking that extracts visual features to re-identify and track a specific suspect across different camera feeds or disconnected frames.
• Hardware Acceleration via DirectML (test_directml.py): Experimental support for Microsoft DirectML, allowing the system to utilize GPU acceleration across diverse hardware (AMD, Intel, NVIDIA) on Windows.
• Automated Security Logging: All security events, identified threats, and anomalies are automatically logged with timestamps and metadata into security_logs.csv and the analytics_output/ directory for post-event analysis.

---

🛠️ Frameworks & Technologies Used

• Computer Vision & AI:
  • Ultralytics YOLOv8 - For real-time object detection and tracking.
  • OpenCV (cv2) - For video capture, frame processing, and image manipulations.
  • PyTorch / Torchvision - Backend for deep learning feature extraction and model inference.
• Hardware Acceleration:
  • DirectML - Extends hardware-accelerated AI to a broader range of GPUs via onnxruntime-directml or PyTorch-DirectML.
• Data Handling & Analytics:
  • Pandas / CSV - For logging security events and handling structured data.
  • NumPy / SciPy - For calculating cosine similarities and spatial distances in the ReID modules.
• Dashboard & Orchestration:
  • Python's standard OS/Subprocess or GUI libraries (Tkinter/PyQt/Streamlit) utilized in main.py to route and execute the modular scripts.

---

📂 Project Architecture

SSF-Vision/
│
├── main.py                  # The main dashboard & entry point to the system
├── anomaly.py               # Core engine for detecting anomalous events
├── suspect_finder.py        # Logic to locate a specific individual in historical/live feeds
├── reid.py / reid1.0.py     # Person Re-Identification (ReID) implementation
├── surveillance_emy.py      # Camera/Feed-specific surveillance configuration 1
├── surveillance_nashwa.py   # Camera/Feed-specific surveillance configuration 2
├── test_directml.py         # Hardware acceleration diagnostics script
├── yolov8n.pt               # Pre-trained YOLOv8 Nano model weights
├── security_logs.csv        # Database of automated security alerts and event logs
├── requirements.txt         # Project dependencies
├── model/                   # Directory containing additional specific ML models
└── analytics_output/        # Directory for generated analytics, snapshots, and reports

---

⚙️ Installation & Setup

1. Clone the repository:

git clone [https://github.com/Mr-Joseph-Jo/SSF-Vision.git](https://github.com/Mr-Joseph-Jo/SSF-Vision.git)
cd SSF-Vision

2. Create a Virtual Environment (Recommended):

python -m venv venv
# On Windows:
venv\Scripts\activate
# On macOS/Linux:
source venv/bin/activate

3. Install Dependencies:

pip install -r requirements.txt

(Note: If you plan to use DirectML on Windows, ensure you install the specific torch-directml or onnxruntime-directml packages as required by your environment).

---

💻 Usage

Starting the Dashboard

To start the primary dashboard which controls the entire surveillance suite, run:

python main.py

From the dashboard, you can route to individual feeds, initialize the ReID system, or toggle anomaly detection.

Running Modules Independently

If you prefer to run specific modules directly from the CLI for testing:

• Test GPU/DirectML Capabilities:

python test_directml.py

• Run Anomaly Detection System:

python anomaly.py

• Run Suspect Search/ReID:

python suspect_finder.py

---

📊 Analytics & Event Logging

SSF-Vision relies heavily on accountability and auditing.

• Any time a rule is broken or a suspect is flagged, a record is appended to security_logs.csv.
• Visual proof (such as bounding box snapshots or tracking paths) is saved in the analytics_output/ folder. This allows security personnel to review exact timestamps and visual evidence of events without watching hours of footage.

---

🤝 Contributing

Contributions are always welcome! If you have suggestions for improving model accuracy, optimizing the dashboard, or adding new deployment methods (like Docker), please fork the repository and create a pull request.

For major changes, please open an issue first to discuss what you would like to change.

📄 License

This system relies on open-source frameworks. Please review the licenses of YOLOv8 (AGPL-3.0) and other utilized libraries. Ensure you comply with all local privacy, surveillance, and data-protection laws (e.g., GDPR) prior to deploying this software in private or public spaces.