git clone https://github.com/Phoenix8215/FCN_KITTI # clone
cd
pip install -r requirements.txt # install-
use
make_tsv.pyto generate the train set,validation set and test set withtsvformat,the set of validation and testing is 80%,20% respectively. -
backbone use VGG16,loss function is binary cross entropy,training 60 epochs.
-
use
python3 train.pyto train the model,and visual the process of training.
- use
python3 test.pyto predict ,outputs are stored inoutputfolder.
- generated weight files which is in
ckptfolder.





