• RSFMDet: Foundation Model-based Auxiliary Framework for Object Detection in Aerial Remote Sensing Images
  • Table of Contents
  • Introduction
  • Installation
  • Training
  • Testing
  • License
  • Citation
  • Acknowledgements

RSFMDet is a foundation model-based auxiliary framework for object detectionbased on MMDetection, designed for efficient and accurate detection in aerial and satellite imagery.

1. Clone this repository:

git clone https://github.com/LZS1991/RSFMDet.git
cd RSFMDet

2. Install dependencies:

pip install -r requirements.txt

3. Install MMDetection (if not already installed):

pip install mmdet

For detailed installation instructions, please refer to the MMDetection documentation.

To train a model with a specific configuration:

python tools/train.py configs/_lzs2851_dior/retinanet_convnextv2_atto_fpn_3x_coco.py

For distributed training:

bash tools/dist_train.sh configs/retinanet_convnextv2_atto_fpn_3x_coco.py 4

To evaluate a trained model:

python tools/test.py configs/_lzs2851_dior/retinanet_convnextv2_atto_fpn_3x_coco.py work_dirs/epoch_50.pth

For distributed testing:

bash tools/dist_test.sh configs/_lzs2851_dior/retinanet_convnextv2_atto_fpn_3x_coco.py work_dirs/epoch_50.pth 4

This project is licensed under the MIT License - see the LICENSE file for details.

If you find RSFMDet useful in your research, please consider citing:

@article{rsfmdet2025,
  title={Foundation Model-based Auxiliary Framework for Object Detection in Aerial Remote Sensing Images},
  author={Wanjie Lu, Chaoyang Niu, Wei Liu, Tao Hu, Chaozhen Lan, and Shiju Wang},
  journal={<Journal Name>},
  year={2025}
}

This project is built upon the excellent MMDetection framework. We thank the MMDetection team for their great work.