irvingao / ganet

A Keypoint-based Global Association Network for Lane Detection. Accepted by CVPR 2022

GANet

This repo is the PyTorch implementation for our paper:

A Keypoint-based Global Association Network for Lane Detection. Accepted by CVPR 2022.

In this paper, we propose a Global Association Network (GANet) to formulate the lane detection problem from a new perspective, where each keypoint is directly regressed to the starting point of the lane line instead of point-by-point extension. Concretely, the association of keypoints to their belonged lane line is conducted by predicting their offsets to the corresponding starting points of lanes globally without dependence on each other, which could be done in parallel to greatly improve efficiency. In addition, we further propose a Lane-aware Feature Aggregator (LFA), which adaptively captures the local correlations between adjacent keypoints to supplement local information to the global association.

Installation

1. Create a conda virtual environment and activate it.

   conda create -n ganet python=3.7 -y
   conda activate ganet
   conda install pytorch==1.6.0 torchvision==0.7.0 cudatoolkit=10.1 -c pytorch -y
   pip install -r requirements/build.txt

2. Clone this repository and enter it:

   git clone https://github.com/Wolfwjs/GANet.git
   cd GANet
   python setup.py develop

Dataset

Download datasets and put it into [data-path] folder. And edit the data_root in the config file to your dataset path.

CULane

[CULane website] [Download]

The directory structure should be like follows:

[data-path]/culane
├── driver_23_30frame
├── driver_37_30frame
├── driver_100_30frame
├── driver_161_90frame
├── driver_182_30frame
├── driver_193_90frame
└── list
    └── test_split
    |   ├── test0_normal.txt
    |   ├── test1_crowd.txt
    |   ├── test2_hlight.txt
    |   ├── test3_shadow.txt
    |   ├── test4_noline.txt
    |   ├── test5_arrow.txt
    |   ├── test6_curve.txt
    |   ├── test7_cross.txt
    |   └── test8_night.txt
    └── train.txt
    └── test.txt
    └── val.txt

TuSimple

[TuSimple website] [Download]

The directory structure should be like follows:

[data-path]/tusimple
├── clips
├── label_data_0313.json
├── label_data_0531.json
├── label_data_0601.json
├── label_data_0601.json
├── test_label.json
└── test_baseline.json

Evaluation

Here are our GANet models (released on April 24th, 2022):

CULane

Version	Backbone	FPS	F1	Weights
Small	ResNet18	153	78.79	ganet_culane_resnet18.pth
Medium	ResNet-34	127	79.39	ganet_culane_resnet34.pth
Large	ResNet-101	63	79.63	ganet_culane_resnet101.pth

TuSimple

Version	Backbone	FPS	F1	Weights
Small	ResNet18	153	97.71	ganet_tusimple_resnet18.pth
Medium	ResNet-34	127	97.68	ganet_tusimple_resnet34.pth
Large	ResNet-101	33	97.45	ganet_tusimple_resnet101.pth

To evalute the model, download the corresponding weights file into the [CHECKPOINT] directory and run the following commands.

# For example, model = ganet-small 
sh slurm_test.sh [PARTITION] [JOB_NAME] tusimple final_exp_res18_s8 [CHECKPOINT]
sh dist_test.sh tusimple final_exp_res18_s8 [CHECKPOINT]

We use the official evaluation tools of CULane and TuSimple to evaluate the results. And we include them in tools directory which may be helpful for you.

Training

# For example, model = ganet-small 
sh slurm_train.sh [PARTITION] [JOB_NAME] tusimple final_exp_res18_s8 ./output
sh dist_train.sh tusimple final_exp_res18_s8 ./output

Citation

If you find this repo useful for your research, please cite

@inproceedings{ganet-cvpr2022,
  author    = {Jinsheng Wang, Yinchao Ma, Shaofei Huang, Tianrui Hui, Fei Wang, Chen Qian, Tianzhu Zhang},
  title     = {A Keypoint-based Global Association Network for Lane Detection},
  booktitle = {CVPR},
  year      = {2022},
}

Contact

For questions about our paper or code, please contact Jinsheng Wang or Yinchao Ma