/SAN

Exploring Self-attention for Image Recognition, CVPR2020.

Primary LanguagePythonMIT LicenseMIT

Exploring Self-attention for Image Recognition

by Hengshuang Zhao, Jiaya Jia, and Vladlen Koltun, details are in paper.

Introduction

This repository is build for the proposed self-attention network (SAN), which contains full training and testing code. The implementation of SA module with optimized CUDA kernels are also included.

Usage

  1. Requirement:

    • Hardware: tested with 8 x Quadro RTX 6000 (24G).
    • Software: tested with PyTorch 1.4.0, Python3.7, CUDA 10.1, CuPy 10.1, tensorboardX.
  2. Clone the repository:

    git clone https://github.com/hszhao/SAN.git
  3. Train:

    • Download and prepare the ImageNet dataset (ILSVRC2012) and symlink the path to it as follows (you can alternatively modify the relevant path specified in folder config):

      cd SAN
      mkdir -p dataset
      ln -s /path_to_ILSVRC2012_dataset dataset/ILSVRC2012
      
    • Specify the gpus (usually 8 gpus are adopted) used in config and then do training:

      sh tool/train.sh imagenet san10_pairwise
      
    • If you are using SLURM for nodes manager, uncomment lines in train.sh and then do training:

      sbatch tool/train.sh imagenet san10_pairwise
  4. Test:

    • Download trained SAN models and put them under folder specified in config or modify the specified paths, and then do testing:

      sh tool/test.sh imagenet san10_pairwise
  5. Visualization:

    • tensorboardX incorporated for better visualization regarding curves:

      tensorboard --logdir=exp/imagenet
  6. Other:

    • Resources: GoogleDrive LINK contains shared models.

Performance

Train Parameters: train_gpus(8), batch_size(256), epochs(100), base_lr(0.1), lr_scheduler(cosine), label_smoothing(0.1), momentum(0.9), weight_decay(1e-4).

Overall result:

Method top-1 top-5 Params Flops
ResNet26 73.6 91.7 13.7M 2.4G
SAN10-pair. 74.9 92.1 10.5M 2.2G
SAN10-patch. 77.1 93.5 11.8M 1.9G
ResNet38 76.0 93.0 19.6M 3.2G
SAN15-pair. 76.6 93.1 14.1M 3.0G
SAN15-patch. 78.0 93.9 16.2M 2.6G
ResNet50 76.9 93.5 25.6M 4.1G
SAN19-pair. 76.9 93.4 17.6M 3.8G
SAN19-patch. 78.2 93.9 20.5M 3.3G

Citation

If you find the code or trained models useful, please consider citing:

@inproceedings{zhao2020san,
  title={Exploring Self-attention for Image Recognition},
  author={Zhao, Hengshuang and Jia, Jiaya and Koltun, Vladlen},
  booktitle={CVPR},
  year={2020}
}