Train and test BiseNet architecure on CamVid to get an understanding of new semantic segmentation architectures.
Obtain camvid dataset from google drive https://drive.google.com/file/d/1KRRME_NtRG-iWOyLAb7gE-eA8fTeyzUR/view
If you are interested in just running the demo,please mention the path to best_dice_loss.pth(it's currently stored in models)
###The architecture was trained on colab for 100 epochs with each epoch taking 2.3 minutes ,suggestions on improving the training process are welcome.
%load_ext tensorboard %tensorboard --logdir runs/