Awesome

Asym-Siam: On the Importance of Asymmetry for Siamese Representation Learning

This is a PyTorch implementation of the Asym-Siam paper, CVPR 2022:

@inproceedings{wang2022asym,
  title     = {On the Importance of Asymmetry for Siamese Representation Learning},
  author    = {Xiao Wang and Haoqi Fan and Yuandong Tian and Daisuke Kihara and Xinlei Chen},
  booktitle = {IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)},
  year      = {2022}
}

The pre-training code is built on MoCo, with additional designs described and analyzed in the paper.

The linear classification code is from SimSiam, which uses LARS optimizer.

Installation

1. Install git

2. Install PyTorch and ImageNet dataset following the official PyTorch ImageNet training code.

3. Install apex for the LARS optimizer used in linear classification. If you find it hard to install apex, it suffices to just copy the code directly for use.

4. Clone the repository:

git clone https://github.com/facebookresearch/asym-siam & cd asym-siam

1 Unsupervised Training

This implementation only supports multi-gpu, DistributedDataParallel training, which is faster and simpler; single-gpu or DataParallel training is not supported.

1.1 Our MoCo Baseline (BN in projector MLP)

To do unsupervised pre-training of a ResNet-50 model on ImageNet in an 8-gpu machine, run:

python main_moco.py \
  -a resnet50 \
  --lr 0.03 \
  --batch-size 256 \
  --dist-url 'tcp://localhost:10001' --multiprocessing-distributed --world-size 1 --rank 0 \
  [your imagenet-folder with train and val folders]

This script uses all the default hyper-parameters as described in the MoCo v2 paper. We only upgrade the projector to a MLP with BN layer.

1.2 MoCo + MultiCrop

python main_moco.py \
  -a resnet50 \
  --lr 0.03 \
  --batch-size 256 \
  --dist-url 'tcp://localhost:10001' --multiprocessing-distributed --world-size 1 --rank 0 \
  [your imagenet-folder with train and val folders] --enable-multicrop

By simply setting --enable-multicrop to true, we can have asym MultiCrop on source side.

1.3 MoCo + ScaleMix

python main_moco.py \
  -a resnet50 \
  --lr 0.03 \
  --batch-size 256 \
  --dist-url 'tcp://localhost:10001' --multiprocessing-distributed --world-size 1 --rank 0 \
  [your imagenet-folder with train and val folders] --enable-scalemix

By simply setting --enable-scalemix to true, we can have asym ScaleMix on source side.

1.4 MoCo + AsymAug

python main_moco.py \
  -a resnet50 \
  --lr 0.03 \
  --batch-size 256 \
  --dist-url 'tcp://localhost:10001' --multiprocessing-distributed --world-size 1 --rank 0 \
  [your imagenet-folder with train and val folders] --enable-asymm-aug

By simply setting --enable-asymm-aug to true, we can have Stronger Augmentation on source side and Weaker Augmentation on target side.

1.5 MoCo + AsymBN

python main_moco.py \
  -a resnet50 \
  --lr 0.03 \
  --batch-size 256 \
  --dist-url 'tcp://localhost:10001' --multiprocessing-distributed --world-size 1 --rank 0 \
  [your imagenet-folder with train and val folders] --enable-asym-bn

By simply setting --enable-asym-bn to true, we can have asym BN on target side (sync BN for target).

1.6 MoCo + MeanEnc

python main_moco.py \
  -a resnet50 \
  --lr 0.03 \
  --batch-size 256 \
  --dist-url 'tcp://localhost:10001' --multiprocessing-distributed --world-size 1 --rank 0 \
  [your imagenet-folder with train and val folders] --enable-mean-encoding

By simply setting --enable-mean-encoding to true, we can have MeanEnc on target side.

2 Linear Classification

With a pre-trained model, to train a supervised linear classifier on frozen features/weights, run:

python main_lincls.py \
  -a resnet50 \
  --lars \
  --dist-url 'tcp://localhost:10001' --multiprocessing-distributed --world-size 1 --rank 0 \
  --pretrained [your checkpoint path] \
  [your imagenet-folder with train and val folders]

Linear classification results on ImageNet using this repo with 8 NVIDIA V100 GPUs :

License

This project is under the CC-BY-NC 4.0 license. See LICENSE for details.