TraffiX.ai

TraffiX.ai is a Platform for Evaluating and Sharing Traffic Forecasting Datasets and Models

Get Started

Tutorials

Tutorials of deep learning-based traffic forecasting.

Tutorials

Benchmarks

Benchmarks of traffic forecating models.

Benchmarks

GitHub

GitHub repositories of the tutorials and source code.

GitHub

TraffiX.ai

It Provides

Performance of deep learning based traffic forecasting models as Benchmark

Note: This platform is under development. It will be updated frequently in the coming month.

Other Resources

  • Open-source code of tested models
  • Tutorials on traffic forecasting
  • Online runnable code
  • Network-wide traffic state datasets

Our History

In order to speed up the progress of transportation AI research, we built a Transportation Artificial Intelligence Platform to provide deep learning algorithms and runnable environments to public users for solving complicated transportation problems. We built the first two versions of the Transportation AI Platform with the capability of evaluating state-of-the-art deep learning models with no need to implement algorithms from scratch. As time goes by, the flexibility and the reproducibility of deep learning models are becoming more important for facilitating the research community, and thus, we developed the TraffiX.ai platform.

  1. The first Transportation AI Platform is a Java EE platform with the key deep learning modules developed based on deeplearning4j.
  2. The second Version is developed based a MVC architecture and the platform’s key deep learning modules are developed based on PyTorch. [Doc:📰]
  3. The TraffiX.ai is the third version, which provides tutorials via nbviewer, open-source code via GitHub, online code runner via CoLab, and performance benchmarks of deep learning-based traffic forecasting algorithms implemented by PyTorch.

Supported By