Project Overview

This project aimed to use a machine learning model to provide a reliable fire and smoke detection system. The YOLOv8 algorithm detects fire and smoke in real time from a video stream. When either is detected, the system triggers sound alerts and sends email notifications with captured screenshots.

Features

• Real-time Detection: The system utilizes a YOLOv8 model trained to detect fire and smoke in real-time via a connected camera.

• Email Alerts: When fire or smoke is detected, the system sends an automated email with a screenshot of the event.

• Sound Notifications: In addition to email alerts, the system plays an alert sound when fire or smoke is detected.

• Customizable Confidence Thresholds: The detection sensitivity for fire and smoke can be adjusted based on specific confidence thresholds to reduce false positives.

• Screenshot Logging: Screenshots of fire or smoke detections are saved automatically with a timestamp for logging purposes.

Requirements

To run this project, you will need to install the following dependencies:

• Python 3.12 or higher
• YOLOv8 (pip install ultralytics)
• OpenCV (pip install opencv-python)
• PyGame (pip install pygame)
• smtplib (Python's built-in email library, no installation needed)
• EmailMessage (from email.message module, built into Python)

You also need to download the trained model (optimized150.pt) and place it in the project directory.

Installation

Follow these steps to set up and run the project:

1. Clone the Repository:

   git clone https://github.com/TalhaKarakoyunlu/Fire-and-Smoke-Detection.git
   cd Fire-and-Smoke-Detection
   

2. Install the Dependencies:

   pip install -r requirements.txt
   

3. Download YOLOv8 Model: Make sure to place the latest trained model (optimized150.pt) in the project directory.

4. Configure Email Credentials: Replace your_email and your_password in the send_email() function.

5. Prepare Alert Sound: Ensure that the alert_sound.mp3 file is in the project directory, you can also use an alert sound that you like.

How to Run

1. Run the YOLOv8 Live Detection Script:

   python YOLOv8LiveCam.py
   

2. Fire and Smoke Detection Logic:

   • Fire Detection: If a fire is detected with confidence >0.55 for 2 seconds continuously, a screenshot is saved, an email is sent, and an alert sound plays.
   • Smoke Detection: If smoke is detected with confidence >0.75 for 3 seconds continuously, a screenshot is saved, an email is sent, and an alert sound plays.

3. Exit: Press the d key to exit the detection loop and close the webcam feed or use CTRL+C to keyboard interrupt.

Customization

1. Adjust Detection Thresholds: You can modify the detection duration threshold for fire and smoke:

   detection_threshold_fire = 2  # Number of seconds fire must be detected
   detection_threshold_smoke = 3  # Number of seconds smoke must be detected
   

2. Change Confidence Levels: Modify the confidence thresholds for fire and smoke detection:

   if class_index == 0 and confidence > 0.55:  # Fire detection threshold
   if class_index == 1 and confidence > 0.75:  # Smoke detection threshold
   

Troubleshooting

• Video Source Error: If there’s an error like "Error: Could not open video source," try changing the camera index in cv2.VideoCapture():

  capture = cv2.VideoCapture(1)  # Change the index to 0 for your integral cam, 2 or 3 for other cameras
  

• Email Issues: Ensure your Gmail is properly configured for SMTP, using an App Password or enabling less secure app access.

• Alert Sound Issues: Ensure alert_sound.mp3 is present in the project directory and pygame is correctly installed.

Extra

• YOLOv8.py can be used to test the models on images and videos.

License

This project is licensed under the MIT License.