BiSeNet based on pytorch 0.4.1 and python 3.6

Download CamVid dataset from Google Drive or Baidu Yun(6xw4).

Download best_dice_loss_miou_0.655.pth in Google Drive or in Baidu Yun(6y3e) and put it in ./checkpoints

python demo.py

Original	GT	Predict

python train.py

Use tensorboard to see the real-time loss and accuracy

python test.py

This time I train the model with dice loss and get better result than cross entropy loss. I did not use lots special training strategy, you can get much better result than this repo if using task-specific strategy.
This repo is mainly for proving the effeciveness of the model.
I also tried some simplified version of bisenet but it seems does not preform very well in CamVid dataset.

This shows the speed comparison between paper and my implementation.

1. The number in first row means input image resolution.
2. The number in second and third row means FPS.
3. The result is based on resnet-18.

• Finish real-time segmentation with camera or pre-load video

• Semantic-Segmentation-Suite
• BiSeNet-paper