ADD-GCN: Attention-Driven Dynamic Graph Convolutional Network for Multi-Label Image Recognition

This project hosts the code for implementing the ADD-GCN algorithm for multi-label image recognition, as presented in our paper:

Attention-Driven Dynamic Graph Convolutional Network for Multi-Label Image Recognition;
Jin Ye, Junjun He, Xiaojiang Peng, Wenhao Wu, Yu Qiao;
In: European Conference on Computer Vision (ECCV), 2020.
arXiv preprint arXiv:2012.02994

The full paper is available at: https://arxiv.org/abs/2012.02994.

Installation

This project is implemented with Pytorch and has been tested on version Pytorch 1.0/1.1/1.2.

A quick demo

After you have installed Pytorch, you can follow the below steps to run a quick demo.

Inference for COCO2014

python main.py --data COCO2014 --data_root_dir {YOUR-ROOT-DATA-DIR} --model_name ADD_GCN --resume {THE-TEST-MODEL} -e -i 448

Please note that:

You should put the COCO2014 folder in {YOUR-ROOT-DATA-DIR}.
You should put the test model in {THE-TEST-MODEL} folder.
You can get the same ADD-GCN results with this model. The password is 4ebj.

Model	Test size	mAP
ResNet-101	448×448	79.7
DecoupleNet	448×448	82.2
ML-GCN	448×448	83.0
ADD-GCN	448×448	84.2
ResNet-101	576×576	80.0
SSGRL	576×576	84.2
ML-GCN	576×576	84.3
ADD-GCN	576×576	85.2

Citations

Please consider citing our paper in your publications if the project helps your research. BibTeX reference is as follows.

@inproceedings{ye2020add,
  title   =  {Attention-Driven Dynamic Graph Convolutional Network for Multi-Label Image Recognition},
  author  =  {Jin Ye, Junjun He, Xiaojiang Peng, Wenhao Wu, Yu Qiao},
  booktitle =  {European Conference on Computer Vision (ECCV)},
  year    =  {2020}
}