This project is a web application that connects to an RTSP live video stream, displays it on the frontend, and analyzes the last frame to detect the presence of smoke and/or fire. The results, including probability scores for each category, are displayed in real-time on the frontend.

• Real-time video streaming from an RTSP or HTTP (JPG/MJPEG) source.
• Fire and smoke detection using a pre-trained deep learning model.
• Real-time display of detection results with probability scores.
• User-configurable threshold for highlighting detection results.
• User-configurable analysis framerate.
• Webhook Notifications: Send alerts to a configurable webhook URL (e.g., Discord) upon detection, including detection scores and the triggering image frame.

1. Clone the repository:

   git clone <repository-url>
   cd <repository-name>

2. Install dependencies:

   pip install -r backend/requirements.txt

3. Configure Webhook (Optional): Create a .env file in the project root and add your webhook URL:

   WEBHOOK_URL=https://discord.com/api/webhooks/your-webhook-id/your-webhook-token
   

   Replace with your actual Discord webhook URL or any other webhook endpoint.

1. Start the backend server:

   uvicorn backend.main:app --reload --port 5000

2. Open the frontend: Open localhost:5000 in your web browser.

This is the recommended way to run the application on a server.

1. Prerequisites:

   • Docker and Docker Compose must be installed on your server.

2. Create .env file: If you haven't already, create a .env file in the project root. This is where you'll put your webhook URL.

   nano .env

   Add your webhook URL:

   WEBHOOK_URL=https://discord.com/api/webhooks/your-webhook-url
   

3. Build and Run: From the project root, run the following command:

   docker-compose up --build -d

   The -d flag runs the application in the background.

4. Access the Application: Open a web browser and navigate to http://<your_server_ip>:5000.

5. To Stop:

   docker-compose down

• The live video stream will be displayed on the page.
• The detection results (Normal, Smoke, Fire) will be displayed with their probability scores in percentages.
• You can adjust the detection threshold using the slider. Categories with scores above the threshold will be highlighted (Fire/Smoke in red/grey, Normal in green).
• If configured, webhook notifications will be sent to your specified URL upon detection.

• You can use the built-in demo stream suggestions for quick testing.
• Here is a screenshot of the application in action: