Car Parking Space Detection using YOLO

This is a computer vision project that uses YOLO (You Only Look Once) algorithm to detect the availability of parking spaces in a parking lot in real-time.

Project Overview

The goal of this project is to develop a system that can detect the number of available parking spaces in real-time, which can be used to optimize parking lot management and improve user experience. The project consists of the following components:

• Dataset: A dataset of parking lot images with labeled parking spaces.
• Preprocessing: Normalization of the dataset to improve the accuracy and performance of the YOLO algorithm.
• Training: Training the YOLO model on the preprocessed dataset using a deep learning framework such as Darknet or TensorFlow.
• Testing and Evaluation: Evaluating the performance of the model using metrics such as mean average precision (mAP) and intersection over union (IoU).
• Real-time Detection: Deploying the model for real-time detection of parking spaces in a video stream or a camera feed.

Installation

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

• Python 3
• OpenCV
• NumPy
• YOLO weights and
• configuration files

You can install OpenCV and NumPy using pip:

pip install opencv-python
pip install numpy

To download the YOLO weights and configuration files, please follow the instructions in this link.

Usage

To run the car parking space detection system, you can use the following command:

python detect_parking.py --input [input video file or camera index] --output [output video file] 

For example:

 python detect_parking.py --input 0 --output output.avi

This will detect parking spaces in the camera feed from camera index 0 and save the output video to a file named output.avi.

Credits

This project was inspired by this article by Carlo Ricci.

License

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