Skip to content
Adversarial Complementary Learning for Weakly Supervised Object Localization
Branch: master
Clone or download
Permalink
Type Name Latest commit message Commit time
Failed to load latest commit information.
datalist add train list and scripts for cub Aug 15, 2018
exper fix the bug of reading images Aug 15, 2018
figs add revisiting.png Jul 31, 2018
models fix the bug of reading images Aug 15, 2018
scripts fix the bug of reading images Aug 15, 2018
utils fix the bug of reading images Aug 15, 2018
LICENSE Initial commit Jul 5, 2018
README.md Update README.md Aug 15, 2018
__init__.py first commit Jul 5, 2018

README.md

Adversarial Complementary Learning for Weakly Supervised Object Localization

Revisiting CAM

We prove the CAM method can be simplified to enable end-to-end training. The proof refers to Section 3.1.

The proposed ACoL method

We apply two classifiers to discover complementary regions of target objects.

Localization

Effect of mining complementary regions

Prerequisites

  • Python2.7
  • PyTorch
  • tqdm

Data Preparation

  • Download the ILSVRC dataset and save them to $data$

Train

git clone https://github.com/xiaomengyc/ACoL.git
cd ACoL
mkdir snapshots
cd scripts
bash train_vgg_imagenet.sh

Citation

If you find this code helpful, please consider to cite this paper:

@inproceedings{zhang2018adversarial,
  title={Adversarial complementary learning for weakly supervised object localization},
  author={Zhang, Xiaolin and Wei, Yunchao and Feng, Jiashi and Yang, Yi and Huang, Thomas},
  booktitle={IEEE CVPR},
  year={2018}
}
You can’t perform that action at this time.