Sampling-Balance_Multi-stage_Network

This code is to replicate the expeirments from the paper Improving Training Instance Quality in Aerial Image Object Detection with A Sampling-balance based Multi-stage Network

1. Introduction

This network is built based on Pytorch 1.1 and MMdetection v1.0rc1

Please refer to INSTALL.md to install the MMdetection framework. It should be noted that a correct version of MMdetection should be download first, which can ensure the codes to be executable.

You should first prepare the used dataset as VOC format.

Three datasets NWPU-VHR10, DIOR, HRRSD are implemented in the branch. If you use one of them, you can set

dataset_type = 'VOCDataset' # for NWPU10

dataset_type = 'HRRSD_Dataset' # for HRRSD dataset_type = 'DOIR_Dataset' # for DIOR

The codes can repreduce the expeiremnts in the paper.

2. The overall architecture of the proposed detector.

3. Some prediction examples of the proposed method on the NWPU VHR-10 data set (Green boxes are the correct predictions. Blue boxes are the false predictions. Red boxes are the missing predictions).

Citation

If you use this method in your research, please cite this paper.

@article{SBNet,
  title   = {Improving Training Instance Quality in Aerial Image Object 
             Detection With a Sampling-Balance-Based Multistage Network},
  author  = {Wei Han, Runyu Fan, Lizhe Wang, Ruyi Feng, Fengpeng Li, 
             Ze Deng, and Xiaodao Chen},
  journal = {{IEEE} Trans. Geosci. Remote. Sens.}, 
  doi     = {10.1109/TGRS.2020.3038803},
  year    = {2020},
  pages   = {1-15}
}