Parking Spot Detector

Parking spot detection for real parking images using YOLO. This branch includes data processing, training scripts for multiple YOLO versions, video testing, and a small Flask API for image detection.

For the simplified YOLOv8 project using HotWheels cars in a simulated parking environment, see the hotwheels branch.

Setup

git clone https://github.com/marcosschlick/parking-spot-detector.git
cd parking-spot-detector
pip install -r requirements.txt

Dataset

Download the parking spot dataset and place it at the project root:

parking-spot-dataset/

Expected raw data layout:

parking-spot-dataset/raw/images
parking-spot-dataset/raw/annotations

Preprocess Data

Run the processing scripts in order:

python src/data_processing/resize_dataset.py
python src/data_processing/labelme_2_yolo.py
python src/data_processing/organize_dataset.py

The last command creates the YOLO dataset structure in dataset/ and generates dataset/data.yaml.

Train

Use one of the version-specific training scripts:

python src/scripts/train_model_v8.py
python src/scripts/train_model_v9.py
python src/scripts/train_model_v10.py
python src/scripts/train_model_v11.py

Each script uses config.yaml and saves results under a version-specific folder:

results/yolo_v8/result/weights/best.pt
results/yolo_v9/result/weights/best.pt
results/yolo_v10/result/weights/best.pt
results/yolo_v11/result/weights/best.pt

Test With Video

The default test script uses YOLOv8:

python src/scripts/test_latest_model.py

To test another version, change MODEL_VERSION at the top of src/scripts/test_latest_model.py.

API

Start the API server:

python src/api/app.py

Test with the default YOLOv8 model:

curl -X POST -F "image=@images-api-test/image_01.png" http://localhost:5000/detect/image

Test a specific YOLO version:

curl -X POST -F "image=@images-api-test/image_01.png" "http://localhost:5000/detect/image?version=yolov11"

Supported versions are yolov8, yolov9, yolov10, and yolov11.

Project Structure

config.yaml
images-api-test/
LICENSE
parking-spot-dataset/
requirements.txt
src/
  api/
  data_processing/
  scripts/

License

The source code in this project is licensed under the MIT License.

Datasets, third-party dependencies, and pretrained YOLO models are subject to their respective licenses and terms of use.