Project Name: Real-Time Traffic Flow Prediction Using Spatio-Temporal Graph Neural Networks
-
Developed a Real-Time Traffic Prediction System:
- Utilized Graph Neural Networks (GNN) and Graph Convolutional Networks (GCN) to model and predict traffic flow on road networks.
- Focused on leveraging the spatial and temporal characteristics of traffic data for accurate predictions.
-
Implemented a Spatio-Temporal GCN Model:
- Combined GCN layers to capture spatial dependencies between road intersections.
- Incorporated LSTM layers to model temporal dependencies and predict future traffic conditions.
- Trained using PyTorch Geometric to efficiently handle graph-structured data.
-
Simulated and Processed Traffic Data:
- Generated synthetic traffic data on a 5x5 grid network representing road intersections.
- Converted the NetworkX graph and traffic data into a PyTorch Geometric format for model training.
- Achieved real-time prediction by continuously updating the model with new traffic data and visualizing predictions.
Technologies and Tools Used:
-
Python: A programming language for implementation and data manipulation.
)
-
PyTorch Geometric: Library for implementing graph neural networks and handling graph data.
-
NetworkX: Library for creating and manipulating complex networks and graphs.
-
NumPy: Used for numerical computations and handling traffic data arrays.
-
Matplotlib: Visualized traffic data and predictions for real-time monitoring and analysis.
