This repository contains a Python project that uses YOLOv8, Roboflow and OpenCV to detect car parking slot occupancy in real-time. The project is divided into three main parts:
The first part of the project focuses on preparing the data for training the YOLOv8 model. This involves the following steps:
->Collecting images of parking lots that capture various parking scenarios.
->Annotating the collected images to label the occupied and empty parking slots.
->Utilizing Roboflow as a platform for dataset management, annotation, and augmentation.
->Applying data augmentation techniques to enhance the diversity and robustness of the dataset.
->Exporting the prepared dataset in a format compatible with YOLOv8 training, such as YOLO darknet format or COCO format.
The second part of the project focuses on training the YOLOv8 model using the prepared dataset. This part includes the following steps:
->Setting up the training environment by installing the required dependencies, including Python, PyTorch, and other necessary libraries.
->Splitting the dataset into training and validation sets to evaluate the model's performance.
->Configuring the YOLOv8 model by specifying the network architecture, hyperparameters, and other settings.
->Initializing the YOLOv8 model architecture and loading pre-trained weights, if available.
->Training the YOLOv8 model using the training dataset and optimizing its parameters to minimize detection errors.
->Monitoring the training progress and evaluating the model's performance using metrics such as loss, precision, recall, and mean average precision (mAP).
->Fine-tuning the model if necessary by adjusting hyperparameters or training for more epochs.
->Saving the trained YOLOv8 model weights for future use during inference.
The third part of the project involves using the trained YOLOv8 model to detect car parking slot occupancy in real-time. This part includes the following steps:
->Capturing the parking area image and saving it to disk using the capture_parking_area.ipynb notebook.
->Selecting the parking slot coordinates from the captured image using the select slot coordinates.ipynb notebook.
->Running the final_slot_detection.ipynb notebook to detect empty slots in real-time videos and displaying the total number of empty slots in the parking area.
->Utilizing the YOLOv8 model and OpenCV for real-time object detection and post-processing to identify occupied and empty parking slots.
This project provides a valuable learning opportunity for understanding YOLOv8, OpenCV, and real-time object detection. It offers features such as real-time detection of car parking slot occupancy, ease of use, and well-documented code.
We hope you find this project useful and enjoy exploring its capabilities!