Skip to content

R3Det: Refined Single-Stage Detector with Feature Refinement for Rotating Object

License

Notifications You must be signed in to change notification settings

SJTU-Thinklab-Det/r3det-on-mmdetection

 
 

Repository files navigation

R3Det: Refined Single-Stage Detector with Feature Refinement for Rotating Object

Abstract

Techniques:

Pipeline

5

Performance

DOTA1.0

Model Backbone Training data Val data mAP Model Link GPU Image/GPU Anchor Reg. Loss lr schd Data Augmentation Configs
R3Det* ResNet50 600->800 DOTA1.0 trainval DOTA1.0 test 71.90 Google Drive -- Baidu Drive (u8bj) 1X GeForce RTX 2080 Ti 6 H + R smooth L1 2x No r3det_r50_fpn_2x_CustomizeImageSplit.py

R3Det*: R3Det with two refinement stages

Compile

python setup.py install

Train

sh rtools/train.sh

Or equivalent command:

python tools/train.py {configuration-file-path}

Before training, please:

  1. Change the paths in lines 97-98 & 102-103 of dota_image_split.py according to your local DOTA dataset directory.
  2. Run python dota_image_split.py to crop train & val set images into smaller tiles, and generate per-tile label files into the directories you specified in step [1].
  3. Change the lines 4-10 of dotav1_rotational_detection.py. Paths in lines 5-8 shall direct to the folders containing the cropped image tiles and label files generated in step [2].
  4. Have fun with sh rtools/train.sh and watch the model train!

Test

sh rtools/test.sh

Or equivalent command:

python tools/test.py {configuration-file-path} {checkpoint-file-path} --format-only --options submission_dir={path-to-save-submission-files}

Before test, please make sure the checkpoint file path (in rtools/test.sh) is correct.

Citation

If this is useful for your research, please consider cite.

@article{yang2019r3det,
    title={R3Det: Refined Single-Stage Detector with Feature Refinement for Rotating Object},
    author={Yang, Xue et al},
    journal={arXiv preprint arXiv:1908.05612},
    year={2019}
}

@inproceedings{xia2018dota,
    title={DOTA: A large-scale dataset for object detection in aerial images},
    author={Xia, Gui-Song and Bai, Xiang and Ding, Jian and Zhu, Zhen and Belongie, Serge and Luo, Jiebo and Datcu, Mihai and Pelillo, Marcello and Zhang, Liangpei},
    booktitle={Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR)},
    pages={3974--3983},
    year={2018}
}

@article{mmdetection,
  title   = {{MMDetection}: Open MMLab Detection Toolbox and Benchmark},
  author  = {Chen, Kai and Wang, Jiaqi and Pang, Jiangmiao and Cao, Yuhang and
             Xiong, Yu and Li, Xiaoxiao and Sun, Shuyang and Feng, Wansen and
             Liu, Ziwei and Xu, Jiarui and Zhang, Zheng and Cheng, Dazhi and
             Zhu, Chenchen and Cheng, Tianheng and Zhao, Qijie and Li, Buyu and
             Lu, Xin and Zhu, Rui and Wu, Yue and Dai, Jifeng and Wang, Jingdong
             and Shi, Jianping and Ouyang, Wanli and Loy, Chen Change and Lin, Dahua},
  journal= {arXiv preprint arXiv:1906.07155},
  year={2019}
}

Reference

About

R3Det: Refined Single-Stage Detector with Feature Refinement for Rotating Object

Resources

License

Code of conduct

Stars

Watchers

Forks

Packages

No packages published

Languages

  • Python 86.0%
  • Cuda 7.8%
  • C++ 6.1%
  • Other 0.1%