Road-Segmentation

Download and Data format

1. Download dataset from http://www.cvlibs.net/datasets/kitti/eval_road.php or https://drive.google.com/file/d/1gakXvNJmnf9YPBt_taE_rlXduWtvMEqk/view?usp=sharing

2. Unzip zip file

3. Rename the data folder

-- data_road\ 
    -- training\
       -- image\
          -- um_000000.png
          -- um_000001.png
          ...
       -- label\
          -- um_lane_000000.png
          -- um_lane_000001.png
          ...
       -- label_grayscale\
          ...
       -- calib\
          ...
    -- testing\
       -- image\
          -- um_000000.png
       -- calib\
          ...

4. Download weights file from https://drive.google.com/file/d/14bQbLzEjSucD7otv69iC5wCx36mVjDaF/view?usp=sharing

Run script

Run convert.py
Run main.py train 
Run main.py check --ckpt=weights.pth  #predict image

Requirement

python 3.6 up
pytorch (cuda supported)