Reimplemented to be compatiable with modern versions of PyTorch (tested with 1.12.0).
Note that this simplified version only supports 1x1/3x3 kernels with 4/8 orientation channels. Please check the master branch for more information.
cd path_to_the_setup.py
pip install .# Train Baseline
python test/mnist-rot/main.py
# Train ORN
python test/mnist-rot/main.py --use-arf# 1. Use predefined ORN-upgraded models (VGG, ResNet, etc.)
from orn import models
model = models.or_resnet18_align_8()
# Print the full list
print('\n'.join(dir(models)))
# 2. Use the helper function for model conversion
from torchvision import models
from orn import upgrade_to_orn
model = models.resnet18(weights=None)
upgrade_to_orn(model, num_orientation=8, scale_factor=2,
classifier=model.fc, features=model.avgpool, invariant_encoding='align')
print(model)
# 3. Use ORN layers
from orn import ORConv, ORAlign1d, ORPool1dIf you use the code in your research, please cite:
@INPROCEEDINGS{Zhou2017ORN,
author = {Zhou, Yanzhao and Ye, Qixiang and Qiu, Qiang and Jiao, Jianbin},
title = {Oriented Response Networks},
booktitle = {CVPR},
year = {2017}
}