Deepware: An Open-Source Toolkit for Developing and Evaluating Learning-Based and Model-Based Autonomous Driving Models

Deepware is the Deep Learning-based Autonomous Driving Toolkit, mainly focuses on end-to-end and middle-to-middle driving. Deepware is used ROS, Tensorflow, Autoware, Docker and CARLA. Deepware is provided automatic data collector, data extractor, data augmenter, model trainer and driving agents.

• Related works
• Install
• Modules
• Road map
• Data set
• Models
• Paper

Related works

• Related work

Install

• Install

Modules

• Data creation
• Model training
• Model evaluation

Road map

• Road map

Dataset

• Google Drive

Models

• Pre-trained model is available in Google Drive
• It is splitted in model.aa to model.al. the files should be combined to model.zip.

Citing Deepware

If you use Deepware in your work, please cite the paper of IEEE Access:

@ARTICLE{9911616,
  author={Seiya, Shunya and Carballo, Alexander and Takeuchi, Eijiro and Takeda, Kazuya},
  journal={IEEE Access}, 
  title={Deepware: An Open-Source Toolkit for Developing and Evaluating Learning-Based and Model-Based Autonomous Driving Models}, 
  year={2022},
  volume={10},
  number={},
  pages={105734-105743},
  doi={10.1109/ACCESS.2022.3212152}}