Parking Space Detection with Mask R-CNN

This project uses Mask R-CNN to detect available parking spaces in a video stream and sends an SMS alert when a parking space is found. The implementation leverages the pre-trained COCO model for object detection and Twilio for sending SMS notifications.

Requirements

• Python 3.6+
• Libraries:
  • numpy
  • opencv-python
  • twilio
  • mask-rcnn (from the Matterport repository)
• COCO pre-trained weights (will be downloaded automatically if not present)
• Twilio Account (for SMS notifications)

Installation

1. Clone the repository:

   git clone https://github.com/your_username/parking-space-detection.git
   cd parking-space-detection

2. Install required Python packages:

   pip install numpy opencv-python twilio
   pip install git+https://github.com/matterport/Mask_RCNN.git

3. Set up your project directory structure:

   parking-space-detection/
   ├── images/
   │   └── test_images/
   │       └── parking.mp4
   ├── logs/
   ├── mask_rcnn_coco.h5  # This will be downloaded automatically if not present
   ├── your_script.py     # Your Python script file
   

4. Twilio Configuration:

   Obtain the following credentials from your Twilio account:

   • twilio_account_sid
   • twilio_auth_token
   • twilio_phone_number
   • destination_phone_number

   Replace the placeholders in your_script.py with your actual Twilio credentials:

   twilio_account_sid = 'YOUR_TWILIO_SID'
   twilio_auth_token = 'YOUR_TWILIO_AUTH_TOKEN'
   twilio_phone_number = 'YOUR_TWILIO_SOURCE_PHONE_NUMBER'
   destination_phone_number = 'THE_PHONE_NUMBER_TO_TEXT'

Running the Project

1. Ensure the video file parking.mp4 is placed in the correct directory:

   parking-space-detection/
   ├── images/
   │   └── test_images/
   │       └── parking.mp4
   

2. Run the script:

   python your_script.py

3. Exit the script:

   Press 'q' to quit the video display window.

Notes

• GPU Support: If you have a GPU and want to leverage it for faster inference, ensure you have the necessary CUDA and cuDNN libraries installed and modify the Mask R-CNN configuration accordingly.
• Video Source: The script is currently set to process a video file (parking.mp4). To use a webcam instead, set VIDEO_SOURCE = 0 in the script.

Credits

• This project uses the Mask R-CNN implementation by Matterport.
• Twilio is used for sending SMS notifications.

License

This project is licensed under the MIT License. See the LICENSE file for details.