one-shot-texture-segmentation

This is a re-implementation of one shot texture segmentation https://arxiv.org/abs/1807.02654

Part of the code are taken from the original author from https://github.com/ivust/one-shot-texture-segmentation

DTD dataset can be downloaded from https://www.robots.ox.ac.uk/~vgg/data/dtd/

Training step

1. Run generate_texture.py to generate train_texture.npy and val_texture.npy
2. Run train.py

Note

This re-implementation is currently not finished. The model is still underfitting. It might because of the vgg layers should be initialized with pretrained weight. Not planning to update this repo anymore, but anyone interested can carry on this work.