Intelligent Traffic Signal Optimization

Overview

This project aims to improve traffic management by detecting vehicles, optimizing traffic signal timings using a Deep Q-Learning (DQN) model, and evaluating the system's performance. The solution includes three key scripts: main.py for vehicle detection and signal control, accuracy.py for evaluating performance, and training.py for training the DQN model.

File Descriptions

• main.py: Contains the vehicle detection logic using OpenCV. Processes video streams, counts vehicles, and adjusts signal timings using actions from a trained DQN agent.
• accuracy.py: Visualizes the performance of the trained model. Shows optimal time allocation and prediction accuracy for each processed video.
• training.py: Implements the DQN algorithm with a neural network. Trains on vehicle count data and optimizes signal timings by adjusting thresholds. The model is periodically saved during training.

Detailed Script Descriptions

main.py

This script captures video frames, applies background subtraction for vehicle detection, and draws bounding boxes around detected vehicles. It counts vehicles crossing a set line and uses a DQN agent to optimize signal timings based on vehicle counts.

Key functionalities:

• Detects vehicles in video frames using OpenCV.
• Counts vehicles crossing a specified line.
• Uses a pre-trained DQN model to adjust signal timings.
• Visualizes vehicle counts and signal timings on video frames.

accuracy.py

This script evaluates the performance of the traffic signal control system by plotting the optimal time allocation and prediction accuracy.

Key functionalities:

• Plots optimal signal timings over multiple videos.
• Displays prediction accuracy of the DQN model.
• Utilizes Matplotlib for visualization.

training.py

This script trains a Deep Q-Learning model to optimize traffic signal timings. It processes video files, adjusts signal timings based on vehicle counts, and updates the DQN model.

Key functionalities:

• Implements DQN with a neural network.
• Trains the model on simulated traffic scenarios.
• Periodically saves the trained model.
• Adjusts signal timings based on learned policies.

Future Improvements

• Additional Traffic Scenarios: Add different types of vehicles or lane configurations to enhance model robustness.
• Real-Time Deployment: Extend the project to work in a real-time traffic control system with live video feeds.
• Enhanced Reward System: Improve the reward function to better simulate real-world traffic flow behavior.

Acknowledgements

• OpenCV: For the computer vision functionality, such as vehicle detection and background subtraction.
• TensorFlow: For implementing the Deep Q-Learning algorithm.
• Matplotlib: For visualizing the accuracy of predictions and optimal time allocation.

How to Run

Vehicle Detection & Signal Control (main.py)

Run the script with:

python main.py

Accuracy Evaluation & Visualization (accuracy.py)

Run the script with:

python accuracy.py

Model Training (training.py)

Run the script with:

python training.py

By following this structure, the three scripts will allow you to detect vehicles, train the DQN model for optimizing traffic signal timings, and visualize the improvements in traffic management.