Robust Lane Detection via Expanded Self Attention (WACV 2022)

Minhyeok Lee, Junhyeop Lee, Dogyoon Lee, Woojin Kim, Sangwon Hwang, Sangyoun Lee

Overview

This repository is the official PyTorch implementation of Robust Lane Detection via Expanded Self Attention (WACV 2022). Our paper can be found here.

Benchmark Results

Architecture

Results (CULane)

Dataset

Download the CULane dataset.

└── Dataset root/
    ├── annotations_new
    ├── driver_23_30frame
    ├── driver_37_30frame
    ├── driver_100_30frame
    ├── driver_161_90frame
    ├── driver_182_30frame
    ├── driver_193_90frame
    ├── laneseg_label_w16
    ├── laneseg_label_w16_test
    └── list/
        ├── test_split/
        │   ├── test0_normal.txt
        │   ├── test1_crowd.txt
        │   └── ...
        ├── test.txt
        ├── test_gt.txt
        ├── train.txt
        ├── train_gt.txt
        ├── val.txt
        └── val_gt.txt

Training

Edit the config.py before training. Then start training with the following:

python train_mymodel.py

Testing

We provide test code for lane prediction visualization. Modify the best model in config.py Then start testing with the following:

python test.py

Video

Citation

@article{lee2021robust,
  title={Robust lane detection via expanded self attention},
  author={Lee, Minhyeok and Lee, Junhyeop and Lee, Dogyoon and Kim, Woojin and Hwang, Sangwon and Lee, Sangyoun},
  journal={arXiv preprint arXiv:2102.07037},
  year={2021}
}