Sentinel

An advanced AI-powered video surveillance and analysis system that provides automated monitoring, visual Q&A, and real-time video captioning capabilities for CCTV cameras and video feeds.

Features

• Support for multiple video input sources:
  • Local video files
  • Webcam feeds
  • RTSP streams from IP cameras
• Real-time video captioning using Salesforce BLIP model
• Natural language visual Q&A using VILT model
• Interactive Streamlit interface
• Continuous monitoring and analysis

Installation

1. Clone this repository:

git clone https://github.com/yourusername/securade-sentinel.git
cd securade-sentinel

2. Create a virtual environment and activate it:

python -m venv venv
source venv/bin/activate  # On Windows, use: venv\Scripts\activate

3. Install the required dependencies:

pip install -r requirements.txt

Requirements

• Python 3.8+
• streamlit
• torch
• transformers
• opencv-python
• pillow
• numpy

Usage

1. Start the application:

streamlit run app.py

2. Select your video source:

   • Upload a local video file
   • Use your webcam
   • Enter an RTSP stream URL

3. The application will begin processing the video feed and displaying:

   • Live video stream
   • Real-time captions
   • Q&A interface for querying the video content

Models

The application uses two main AI models:

1. Video Captioning: Salesforce/blip-image-captioning-large

   • Generates natural language descriptions of video scenes

2. Visual Q&A: dandelin/vilt-b32-finetuned-vqa

   • Answers questions about the video content in natural language

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.