Paper - FAR: Fourier Aerial Video Recognition (ECCV 2022)

Please cite our paper if you find it useful.

@inproceedings{kothandaraman2022far,
  title={FAR: Fourier Aerial Video Recognition},
  author={Kothandaraman, Divya and Guan, Tianrui and Wang, Xijun and Hu, Shuowen and Lin, Ming and Manocha, Dinesh},
  booktitle={European Conference on Computer Vision},
  pages={657--676},
  year={2022},
  organization={Springer}
}

Code structure

Dataloaders

dataset/dataset.py

Models

model/i3d_resnet.py - I3D model
model/x3d.py - X3D model
model/discaus1.py - I3D + FAR (Ours)
model/x3d_discaus.py - X3D + FAR (Ours)

Dependencies

PyTorch
NumPy
Matplotlib
OpenCV
SciPy

Acknowledgements

This code is heavily borrowed from Benchmarking Action Recognition Models, and X3D-MultiGrid-PyTorch