License

Overview

SafeVision is a comprehensive, multi-modal content safety suite powered by advanced ONNX deep learning models. This sophisticated system provides real-time nudity detection, content filtering, and automated censoring across images, videos, live streams, and camera feeds. With multiple interfaces including CLI tools, modern GUI applications, live monitoring, streaming integration, and RESTful API endpoints, SafeVision offers enterprise-grade content moderation capabilities for various use cases.

🎯 Key Capabilities

Multi-Format Support: Images, videos, live camera feeds, screen capture
Real-time Processing: Live detection with sub-second response times
Multiple Interfaces: CLI, GUI, API server, screen overlay, streaming integration
Advanced AI Models: ONNX-optimized deep learning models with 18+ detection categories
Professional Features: Batch processing, custom rules, alert systems, logging
Cross-Platform: Windows, Linux, macOS support with optimized performance

📑 Table of Contents

⭐ Features

🤖 AI-Powered Detection

Advanced ONNX Models: Optimized deep learning models for accurate content detection
18+ Detection Categories: Comprehensive labeling system for different content types
Risk Assessment: Automatic severity classification (Safe, Low, Moderate, High, Critical)
Real-time Processing: Sub-second analysis with GPU acceleration support
Confidence Scoring: Adjustable detection thresholds for different use cases

🎯 Content Processing

Multi-Format Support: Images (JPG, PNG, BMP, TIFF), Videos (MP4, AVI, MOV, MKV)
Live Camera Feeds: Real-time webcam and USB camera processing
Screen Capture: Monitor desktop activity with overlay detection
Batch Processing: Process multiple files simultaneously
Audio Preservation: Maintain original audio in processed videos

🛡️ Censoring & Safety

Intelligent Blurring: Selective or full-frame blur with adjustable strength
Color Masking: Solid color overlay as alternative to blur
Custom Exception Rules: Define what content to blur or ignore
Protection Modes: Kids Safety, Streamer, Nudity Fighter, and custom modes
Alert Systems: Real-time notifications and emergency actions

🖥️ User Interfaces

Modern PyQt5 GUI: Professional interface with drag-drop, themes, and live preview
Command Line Tools: Full CLI support for automation and scripting
Screen Overlay: Transparent overlay for monitoring any application
REST API Server: HTTP endpoints for integration with other applications
Streaming Integration: OBS and live streaming platform support

📊 Professional Features

Detailed Logging: Comprehensive detection logs with timestamps and metadata
Performance Monitoring: FPS tracking, memory usage, and system optimization
Configuration Management: Persistent settings and customizable parameters
Report Generation: Export detection reports in multiple formats
Multi-threading: Optimized performance with parallel processing

🔧 Development & Integration

RESTful API: Complete API server with JSON responses and file upload support
SDK Components: Reusable classes for custom integration
Event Hooks: Callback system for custom actions on detection
Plugin Architecture: Extensible design for custom detection rules
Cross-platform: Windows, Linux, macOS support with native installers

🔧 Installation & Setup

📋 System Requirements

Minimum Requirements:

Python 3.8+ (3.9+ recommended)
4GB RAM (8GB+ recommended)
2GB free disk space
CPU with AVX2 support (Intel 2013+, AMD 2017+)

Recommended for Optimal Performance:

Python 3.10+
16GB+ RAM
NVIDIA GPU with CUDA support
SSD storage for faster model loading

🚀 Quick Installation

# Clone the repository
git clone https://github.com/im-syn/safevision.git
cd safevision

# Install core dependencies
pip install -r requirements.txt

# For GUI applications (PyQt5 interface)
pip install -r requirements_gui.txt

# For API server functionality
pip install -r requirements_api.txt

# For live streaming features
pip install -r requirements_streaming.txt

📦 Required Models

SafeVision requires ONNX model files in the Models/ directory:

# Create Models directory
mkdir Models

# Place your models (obtain from official source):
# Models/best.onnx           - Main nudity detection model
# Models/best_gender.onnx    - Gender/age detection model (optional)

📥 Model Download: Contact the maintainer or check releases for official model files.

🔧 Advanced Installation Options

GPU Acceleration (Recommended)

# For NVIDIA GPU support
pip install onnxruntime-gpu

# For Intel GPU support  
pip install onnxruntime-openvino

# For AMD GPU support
pip install onnxruntime-directml

FFmpeg Installation (Required for Video Processing)

# Windows (using chocolatey)
choco install ffmpeg

# macOS (using homebrew)
brew install ffmpeg

# Ubuntu/Debian
sudo apt update && sudo apt install ffmpeg

# Manual installation
python download_ffmpeg.py

Virtual Environment (Recommended)

# Create virtual environment
python -m venv safevision_env

# Activate (Windows)
safevision_env\Scripts\activate

# Activate (Linux/macOS)
source safevision_env/bin/activate

# Install dependencies
pip install -r requirements.txt

✅ Installation Verification

# Test core functionality
python main.py --help

# Test GUI interface
python safevision_gui.py

# Test API server
python safevision_api.py

# Test live detection
python live.py --help

# Test video processing
python video.py --help

# Test live streaming
python live_streamer.py --help

🚨 Common Installation Issues

ONNX Runtime Issues

# If you get ONNX import errors:
pip uninstall onnxruntime onnxruntime-gpu
pip install onnxruntime==1.15.1

# For GPU support:
pip install onnxruntime-gpu==1.15.1

OpenCV Issues

# If OpenCV fails to import:
pip uninstall opencv-python opencv-python-headless
pip install opencv-python==4.8.1.78

PyQt5 Issues

# If GUI fails to start:
pip uninstall PyQt5
pip install PyQt5==5.15.9

📁 Project Structure

SafeVision/
├── 🎯 Core Applications
│   ├── main.py                    # Image processing CLI
│   ├── video.py                   # Video processing CLI
│   ├── live.py                    # Live camera detection
│   └── live_streamer.py           # Live streaming integration
│
├── 🎨 User Interfaces
│   ├── safevision_gui.py          # Modern PyQt5 GUI
│   └── safevision_api.py          # REST API server
│
├── 📁 Models & Configuration
│   ├── Models/
│   │   ├── best.onnx             # Main detection model
│   │   └── best_gender.onnx      # Gender/age model (optional)
│   ├── BlurException.rule         # Default blur rules
│   └── custom_rules.rule          # Custom rule examples
│
├── 📁 Output Directories (Auto-created)
│   ├── output/                    # Final processed content
│   ├── Blur/                     # Blurred versions
│   ├── Prosses/                  # Detection visualizations
│   ├── video_output/             # Processed videos
│   ├── Logs/                     # Application logs
│   └── detection_screenshots/    # Detection captures
│
└── 📁 Configuration Files (Auto-generated)
    ├── safevision_settings.json  # GUI settings
    ├── detection_log.json        # Detection history
    └── violation_log.json        # Violation tracking

📄 Key Files Explained

🎯 Core Processing Files

main.py: Single image processing with CLI interface
video.py: Batch video processing with advanced options
live.py: Real-time camera feed analysis
live_streamer.py: Live streaming integration with OBS support

🎨 User Interface Files

safevision_gui.py: Modern PyQt5 interface with advanced features
safevision_api.py: RESTful API server for integration

⚙️ Configuration Files

BlurException.rule: Rules defining what content to blur
*.json: Runtime settings and detection logs
safevision_gui.py: Modern PyQt5 interface with advanced features
safevision_api_server.py: RESTful API server

SafeVision

Install / Use

README