QData / C-Tran

General Multi-label Image Classification with Transformers
Jack Lanchantin, Tianlu Wang, Vicente Ordóñez Román, Yanjun Qi
Conference on Computer Vision and Pattern Recognition (CVPR) 2021
[paper] [poster] [slides]

Training and Running C-Tran

Python version 3.7 is required and all major packages used and their versions are listed in requirements.txt.

C-Tran on COCO80 Dataset

Download COCO data (19G)

wget http://cs.virginia.edu/~jjl5sw/data/vision/coco.tar.gz
mkdir -p data/
tar -xvf coco.tar.gz -C data/

Train New Model

python main.py  --batch_size 16  --lr 0.00001 --optim 'adam' --layers 3  --dataset 'coco' --use_lmt --dataroot data/

C-Tran on VOC20 Dataset

Download VOC2007 data (1.7G)

wget http://cs.virginia.edu/~jjl5sw/data/vision/voc.tar.gz
mkdir -p data/
tar -xvf voc.tar.gz -C data/

Train New Model

python main.py  --batch_size 16  --lr 0.00001 --optim 'adam' --layers 3  --dataset 'voc' --use_lmt --grad_ac_step 2 --dataroot data/

Citing

@article{lanchantin2020general,
  title={General Multi-label Image Classification with Transformers},
  author={Lanchantin, Jack and Wang, Tianlu and Ordonez, Vicente and Qi, Yanjun},
  journal={arXiv preprint arXiv:2011.14027},
  year={2020}
}