TricubeNet - Official Pytorch Implementation (WACV 2022)

TricubeNet: 2D Kernel-Based Object Representation for Weakly-Occluded Oriented Object Detection
Beomyoung Kim¹, Janghyeon Lee², Sihaeng Lee², Doyeon Kim³, Junmo Kim³

¹ _{NAVER CLOVA}
² _{LG AI Research}
³ _KAIST

WACV 2022

How to use?

Data Preparation

DOTA official page

For training

bash run_train.sh

Please check the discription of training hyperparameters (we recommend to use default hyperparameters)

python3 train.py --help

For testing

cd evaluation
bash run_eval.sh

Please check the discription of testing hyperparameters (we recommend to use default hyperparameters)

python3 eval_DOTA.py --help

Qualitative Results

DOTA

MSRA-TD500, ICDAR 2015

SKU110K-R

Citation

We hope that you find this work useful. If you would like to acknowledge us, please, use the following citation:

@inproceedings{kim2022tricubenet,
  title={TricubeNet: 2D Kernel-Based Object Representation for Weakly-Occluded Oriented Object Detection},
  author={Kim, Beomyoung and Lee, Janghyeon and Lee, Sihaeng and Kim, Doyeon and Kim, Junmo},
  booktitle={Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision},
  pages={167--176},
  year={2022}
}

lymhust/TricubeNet