CenLaneNet

This is an implementation of "":

​

Quick starts:

链接: https://pan.baidu.com/s/1lakhm5LCABcMDVoTTzXN8g code: wn1u 1) Evaluating tusimple

python A2_Test_tusimple.py --TusimpleTesting_root "yourTusimplePath/test_set/"

2) Evaluating culane

python A1_Test_culane.py --CUlane_dataroot "/home/sunyi/sy/data/LaneDetection/CULane/"

3) Post-Processing comparison

python A3_Postprocess_cost_test.py --TusimpleTesting_root  "yourTusimplePath/test_set/" --show 1

Training your own model:

1) preparing ground_truth--tusimple

python D0_CreateTusimpleGt.py --tusimple_root "/home/sunyi/sy/data/LaneDetection/tusimple_raw/tosunyi/train_set/" --GtDataroot  "./"

2) Training

python A0_Train.py --GtDataroot "/home/sunyi/sy/data/LaneDetection/tusimple/"

Some parameters are selected from validation datasets... (Code for generating the Gt of Culane will come soon....)

Please contact me:sunyi13@nudt.edu.cn if you have any questions about this implementation。If you find this project is useful, please cite this work if possible: @ARTICLE{9735393, author={Yi, Sun and Li, Jian and Xu, Xin and Shi, Yifei}, journal={IEEE Transactions on Intelligent Vehicles}, title={Adaptive Multi-Lane Detection based on Robust Instance Segmentation for Intelligent Vehicles}, year={2022}, volume={}, number={}, pages={1-1}, doi={10.1109/TIV.2022.3158750}}