FCN for semantic segmentation
This is a simple FCN[1] implementation in Pytorch. There are some differences from the original FCN:
- ResNet features are used
- add dilated convolution
- features of all layers are used
- features of different layers are combined via concatenation instead of summation.
- VOC training set is argumented by flipping and cropping.
I haven't test the performance.
requirement
pytorch, tensorboard-pytorch, tensorboard (for visulization)
If you don't need visulization, then delete the lines about visulization in "main.py".
Usage
Train:
python main.py
Reference
[1] Long, Jonathan, Evan Shelhamer, and Trevor Darrell. "Fully convolutional networks for semantic segmentation." Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. 2015.