This is the official implementation of FCOSF, which is implemented on OBBDetection
- (2023-05-29) Release FCOSF🔥.
- (2023-05-29) Release MMRotate Implementation stay tuning🙍.
Please refer to install.md for installation and dataset preparation.
If you want to train or test a oriented model, please refer to oriented_model_starting.md.
To help you start quickly, I prepare a simple bash script
config=/path/to/config && work_dir=/path/to/work_dir && cuda=(device_id, like 0, 1, 2, 3 ...)
bash my_exps/train.sh ${config} ${work_dir} ${cuda}
config=/path/to/config && ckpt=/path/to/checkpoint && save_dir=/path/to/results_save_dir && cuda=(same as above)
bash my_exps/test.sh ${config} ${ckpt} ${save_dir} ${cuda}
TODO:
@article{fcosf,
author={Rao, Chaofan and Wang, Jiabao and Cheng, Gong and Xie, Xingxing and Han, Junwei},
journal={IEEE Transactions on Geoscience and Remote Sensing},
title={Learning Orientation-aware Distances for Oriented Object Detection},
year={2023},
pages={1-1},
doi={10.1109/TGRS.2023.3278933}}
This project is released under the Apache 2.0 license.