🚗 ParkMatrix AI

Smart Parking Availability Intelligence using Deep Learning

ParkMatrix AI predicts urban parking availability using a CNN–LSTM deep learning model trained on real historical parking data.

---

🚀 Features

• Real-time parking availability prediction
• CNN + LSTM time-series forecasting
• Zone-based urban modeling (Z1 – Z5)
• Interactive Plotly visualizations
• Map-based location intelligence (OpenStreetMap)
• Exact arrival time analysis (12-hour format)
• Best alternative parking time suggestion

---

🧠 Model Architecture

Input: Last 3 Hours Parking Occupancy ↓ Conv1D Layer ↓ LSTM Layer ↓ Dense Layer ↓ Output: Next-Hour Parking Occupancy (%)

Input: Occupancy values of previous 3 hours

• Output: Predicted parking demand (%)
• Framework: TensorFlow / Keras
• Inference: Real-time using trained .keras model

---

🖼️ Screenshots

🔹 Application Interface

---

🔹 Graphical Interpretation

---

🔹 AI Conclusion

---

🔬 Proof This Is a Real Deep Learning Project

• Predictions change when dataset values are modified
• Uses trained cnn_lstm_parking_model.keras
• Requires exactly 3 historical hours for inference
• No hardcoded rules or fake logic
• Model fails gracefully if data is insufficient

---

📊 Tech Stack

• Python
• Streamlit
• TensorFlow / Keras
• Plotly
• Pandas / NumPy
• OpenStreetMap (Nominatim API)

---

▶️ Run Locally

pip install -r requirements.txt
streamlit run app.py


👤 Author
Puneeth Raj Yadav
Aspiring Software Engineer | Deep Learning & Full-Stack Projects