Cross-Domain Weakly-Supervised Object Detection through Progressive Domain Adaptation

This page is for the paper appeared in CVPR2018. You can also find project page for the paper.

Here is the example of our results in watercolor images.

Requirements

• Python 3.5+
• Chainer 3.0+
• ChainerCV 0.8
• Cupy 2.0+
• OpenCV 3+
• Matplotlib

Please install all the libraries. We recommend pip install -r requirements.txt.

Download models

Please go to both models and datasets directory and follow the instructions.

Usage

For more details about arguments, please refer to -h option or the actual codes.

Demo using trained models

python demo.py input/watercolor_142090457.jpg output.jpg --gpu 0 --load models/watercolor_dt_pl_ssd300

Evaluation of trained models

python eval_model.py --root datasets/clipart --data_type clipart --det_type ssd300 --gpu 0 --load models/clipart_dt_pl_ssd300

Training using clean instance-level annotations (ideal case)

python train_model.py --root datasets/clipart --subset train --result result --det_type ssd300 --data_type clipart --gpu 0

Training using virtually created instance-level annotations

Rest of this section shows examples for experiments in clipart dataset.

1. (Preprocess): please follow instructions in ./datasets/README.md to create folders.

2. Domain transfer (DT) step

   1. python train_model.py --root datasets/dt_clipart/VOC2007 --root datasets/dt_clipart/VOC2012 --subset trainval --result result/dt_clipart --det_type ssd300 --data_type clipart --gpu 0 --max_iter 500 --eval_root datasets/clipart

   We provide models obtained in this step at ./models.

3. Pseudo labeling (PL) step

   1. python pseudo_label.py --root datasets/clipart --data_type clipart --det_type ssd300 --gpu 0 --load models/clipart_dt_ssd300 --result datasets/dt_pl_clipart
   2. python train_model.py --root datasets/dt_pl_clipart --subset train --result result/dt_pl_clipart --det_type ssd300 --data_type clipart --gpu 0 --load models/clipart_dt_ssd300 --eval_root datasets/clipart

Citation

If you find this code or dataset useful for your research, please cite our paper:

@inproceedings{inoue2018cross,
  title={Cross-domain weakly-supervised object detection through progressive domain adaptation},
  author={Inoue, Naoto and Furuta, Ryosuke and Yamasaki, Toshihiko and Aizawa, Kiyoharu},
  booktitle={Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition},
  pages={5001--5009},
  year={2018}
}