🚗 Parking Space Occupancy Classification

A machine learning project that classifies parking spaces as Empty or Occupied using image classification and a Support Vector Machine (SVM).

The project demonstrates a complete machine learning workflow, including image preprocessing, dataset preparation, model training, hyperparameter tuning with GridSearchCV, model evaluation, and model serialization.

---

📌 Features

• Binary image classification
• Parking space occupancy prediction
• Image preprocessing and resizing
• Feature extraction using flattened image vectors
• Support Vector Machine (SVM) classifier
• Hyperparameter tuning with GridSearchCV
• Model evaluation using accuracy score
• Trained model serialization with Pickle

---

🛠️ Tech Stack

Category	Technology
Language	Python
Machine Learning	Scikit-learn
Image Processing	scikit-image
Numerical Computing	NumPy
Model	Support Vector Machine (SVM)
Hyperparameter Optimization	GridSearchCV
Model Serialization	Pickle

---

📂 Project Structure

parking-space-occupancy-classification/
│
├── clf-data/
│   ├── empty/
│   └── not_empty/
│
├── main.py
├── model.p
├── requirements.txt
├── .gitignore
└── README.md

---

🔄 Machine Learning Pipeline

Dataset
   │
   ▼
Load Images
   │
   ▼
Resize Images (15 × 15)
   │
   ▼
Flatten Images
   │
   ▼
Train/Test Split
   │
   ▼
GridSearchCV
   │
   ▼
Train SVM Classifier
   │
   ▼
Evaluate Accuracy
   │
   ▼
Save Trained Model

---

📊 Dataset

The dataset contains two classes:

• 🟢 Empty
• 🚗 Not Empty

Each image is resized to 15 × 15 pixels before being converted into a one-dimensional feature vector for training.

---

🚀 Installation

Clone the repository:

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

Install dependencies:

pip install -r requirements.txt

---

▶️ Training

Run the training script:

python main.py

The script will:

• Load the dataset
• Train an SVM classifier
• Search for the best hyperparameters using GridSearchCV
• Evaluate model accuracy
• Save the trained model as model.p

---

📈 Model Evaluation

The model is evaluated using:

• Accuracy Score

---

🔮 Future Improvements

• Deep Learning implementation with PyTorch
• Convolutional Neural Networks (CNNs)
• Real-time parking occupancy detection
• Webcam or CCTV integration
• Web dashboard for monitoring parking availability
• Multi-class parking analysis

---

👨‍💻 Author

Moustafa Serag

GitHub: https://github.com/MoustafaSerag