Backend for MutualDetector
======================


VOC07 and VOC12
-----------------
Step0: install requirements by 'pip install -r requirements.txt'

Step1: set the VOC data path to VOC_ROOT in the file data/voc.py

Step2: run the code
```
Our method: python train_csd.py --dataset VOC12 --model 300 --image_level_supervision True --batch_size 32 --supervise_percent 0.25 --semi True
CSD-SSD: python train_csd.py --dataset VOC12 --model 300 --image_level_supervision True --batch_size 32 --supervise_percent 0 --semi True
```

COCO17
-----------------
To be added


## Citation
If you use this code for your research, please consider citing:
```
@article{chen2022towards,
  title={Towards Better Caption Supervision for Object Detection},
  author={Chen, Changjian and Wu, Jing and Wang, Xiaohan and Xiang, Shouxing and Zhang, Song-Hai and Tang, Qifeng and Liu, Shixia},
  journal={IEEE Transactions on Visualization and Computer Graphics},
  year={2022},
  volume={28},
  number={4},
  pages={1941--1954},
}
```

## Contact
If you have any problem about this code, feel free to contact
- ccj17@mails.tsinghua.edu.cn

or describe your problem in Issues.