Skip to content

Latest commit

 

History

History

ENet-TuSimple-Torch

README.md

Before Start

Please follow list6 and list to put the TuSimple dataset (train, val, test) in the desired folder. We'll call the directory that you cloned ENet-TuSimple-Torch as $ENet_TuSimple_ROOT. Note that if you use ENet-Label-Torch (i.e., ENet-SAD) as the backbone, you can get around 96.64% accuracy in TuSimple testing set.

Testing

  1. Run test script

    cd $ENet_TuSimple_ROOT
sh ./laneExp/ENet-model/test.sh

    Testing results (probability map of lane markings) are saved in predicts/ by default.

  2. Generate json file from probability maps

    python pred_json.py

    The generated json file would be named pred_ENet_test.json by default.

  3. Calculate accuracy, FP and FN

    cd evaluate
python lane.py pred_ENet_test.json label.json

    By now, you should be able to reproduce the result (Accuracy: 0.9486756915, FP: 0.0457901133, FN: 0.0386712197).

Training

  1. Training ENet model 
    cd $ENet_TuSimple_ROOT
sh ./laneExp/ENet-model/train.sh
    The training process should start and trained models would be saved in laneExp/ENet-model/ENet/ by default.
    Then you can test the trained model following the Testing steps above. If your model position or name is changed, remember to set them to yours accordingly.