Are you sure you want to delete this task? Once this task is deleted, it cannot be recovered.
wzhuang 0e5cc99b75 | 1 year ago | |
---|---|---|
.. | ||
Data | 1 year ago | |
Models | 1 year ago | |
DataLoader.py | 1 year ago | |
Demo.bat | 1 year ago | |
F_Conv.py | 1 year ago | |
MyLib.py | 1 year ago | |
README.md | 1 year ago | |
Rotated_MNIST_Main.py | 1 year ago | |
Rotated_MNIST_simpleCase_Main.py | 1 year ago | |
SteerableCNN_XQ.py | 1 year ago |
Folder structure:
Data\ : data folder
|-- mnist_rotation_new.zip : roatated minist data, Unzip before use.
Models\ : Trained result
|-- NormalNet\ : Trained parameters
|-- NormalNet_best_of_30_repetitions\ : Best model parameter of 30 random repetitions
|-- SimpleNet\ : Trained parameters of the simple network in the paper
DataLoader.py : The data reading and preparing code
F_Conv.py : Core codes of the equivariant convolutions of F-Conv
MyLib.py : Some used code
Rotated_MNIST_Main.py : Main code of experiment
Rotated_MNIST_simpleCase_Main.py : Main code of experiment of the simple case in the paper
SteerableCNN_XQ.py : F-Conv based Networks
Demo.bat : Demo for running all the experiments
Usage:
Unzip Data/mnist_rotation_new.zip to get the rotated mnist dataset first, which can also be download from http://www.iro.umontreal.ca/~lisa/icml2007data/mnist_rotation_new.zip.
To run testing with the example data, you can just run Rotated_MNIST_Main.py which is equivariant under reflections call:
python Rotated_MNIST_Main.py --dir NormalNet
To test the best model trained by us, one can call:
python Rotated_MNIST_Main.py --dir NormalNet_best_of_30_repetitions
To train the model, one can call:
python Rotated_MNIST_Main.py --dir retrain --mode train
To test and train the simple cases, one call:
python Rotated_MNIST_simpleCase_Main.py --dir SimpleNet
python Rotated_MNIST_simpleCase_Main.py --dir SimpleNet_Retrain --mode train
Citation:
Qi Xie, Qian Zhao, Zongben Xu and Deyu Meng*.
Fourier Series Expansion Based Filter Parametrization for Equivariant Convolutions[J].
IEEE transactions on pattern analysis and machine intelligence, 2022.
BibTeX:
@article{xie2020MHFnet,
title={Fourier Series Expansion Based Filter Parametrization for Equivariant Convolutions},
author={Xie, Qi and Zhao, Qian and Xu, Zongben and Meng, Deyu},
journal={IEEE Transactions on Pattern Analysis and Machine Intelligence},
year={2022},
publisher={IEEE}
}
Dear OpenI User
Thank you for your continuous support to the Openl Qizhi Community AI Collaboration Platform. In order to protect your usage rights and ensure network security, we updated the Openl Qizhi Community AI Collaboration Platform Usage Agreement in January 2024. The updated agreement specifies that users are prohibited from using intranet penetration tools. After you click "Agree and continue", you can continue to use our services. Thank you for your cooperation and understanding.
For more agreement content, please refer to the《Openl Qizhi Community AI Collaboration Platform Usage Agreement》