/TwinLiteNet

Primary LanguagePythonMIT LicenseMIT

TwinLiteNet: An Efficient and Lightweight Model for Driveable Area and Lane Segmentation in Self-Driving Cars

PWC PWC

Requirement

See requirements.txt for additional dependencies and version requirements.

pip install -r requirements.txt

Data Preparation

  • Download the images from images.

  • Download the annotations of drivable area segmentation from segments.

  • Download the annotations of lane line segmentation from lane.

/data
    bdd100k
        images
            train/
            val/
            test/
        segments
            train/
            val/
        lane
            train/
            val/

Pipeline

Train

python3 main.py

Test

python3 val.py

Inference

Images

python3 test_image.py

Visualize

Drive-able segmentation

Lane Detection

Acknowledgement

Our source code is inspired by:

Citation

If you find our paper and code useful for your research, please consider giving a star ⭐ and citation 📝 :

@misc{che2023twinlitenet,
      title={TwinLiteNet: An Efficient and Lightweight Model for Driveable Area and Lane Segmentation in Self-Driving Cars}, 
      author={Quang Huy Che and Dinh Phuc Nguyen and Minh Quan Pham and Duc Khai Lam},
      year={2023},
      eprint={2307.10705},
      archivePrefix={arXiv},
      primaryClass={cs.CV}
}