Traffic Prediction using Bi-Directional LSTM

AI Traffic Simulation

This repository showcases my work on an AI Traffic Simulation project as part of an Omdena collaboration. The project focused on using AI to optimize urban traffic management.

My Role

• Developed a traffic prediction model using LSTM networks.
• Led Task-3: Traffic Control Algorithm Model Development.
• Conducted daily stand-ups and managed progress using a sprint methodology.

Certificate of Completion

About Omdena

Omdena is a global platform where AI practitioners collaborate to solve real-world challenges.

Project Overview

This project aims to predict traffic conditions, specifically traffic light signals and traffic volume, using a Bi-directional Long Short-Term Memory (LSTM) model. It processes time-series traffic data and trains a multi-task learning model that predicts both traffic light status and traffic volume.

Objectives

The goal of this project is to enhance traffic management systems by providing predictions that can optimize traffic control. The key tasks include:

1. Traffic Light Status: A classification task that predicts the current status of the traffic light (e.g., red, yellow, green).
2. Traffic Volume: A regression task that predicts the number of vehicles on the road.

Approach

This project leverages the power of Bi-directional LSTM (Bi-LSTM) to learn from both past and future traffic data, enhancing the model's ability to capture patterns in traffic behavior. The tasks include:

• Data Collection: Gathering time-series traffic data with relevant features (e.g., timestamps, traffic light status, vehicle count).
• Data Preprocessing: Handling missing data, scaling, and transforming the dataset into a suitable format for training.
• Model Development: Building a Bi-directional LSTM model with two output layers for classification (traffic light status) and regression (traffic volume).
• Model Training: Training the Bi-LSTM model with appropriate loss functions for both tasks.
• Evaluation: Evaluating the model's performance using metrics such as accuracy (for classification) and mean squared error (for regression).

Features

• Multi-Task Learning: Simultaneously predicts both traffic light status and traffic volume using a single model.
• Bi-LSTM Architecture: Captures both forward and backward dependencies in the time-series data.
• Real-time Traffic Predictions: Can be used to predict traffic light status and vehicle count for real-time traffic management.

Video Demonstration

You can view the video demonstration of the traffic management model by clicking the link below:

Watch the video