Pytorch implementation of paper from berkeley@Jonathan Long.
I use KITTI road dataset, you can download link.
You will need the following package before you go through the tutorials.
Recommand you using Anaconda 3 !!!
- pytorch >= 1.0.0
- torchsummary
FCN is using skipped network which stitches the pooling layers of pretrained model to compensate important features while training.
I think everyone eagered to see the result while the beginning. Hence, I recommend you to see what you got after this tutorials. demo.ipynb
Next, you will want to see the structure of this powerful model. introduction.ipynb
Last but not least, you will want to learn how you can build your own model. train.ipynb