BAM & CBAM Pytorch

Pytorch implementation of BAM and CBAM.

BAM & CBAM Pytorch

This code purpose to evaluate of popular attention model architectures, such as BAM, CBAM on the CIFAR dataset.

Park J, Woo S, Lee J Y, Kweon I S. BAM: Bottleneck Attention Module. 2018. BMVC2018(Oral)

Woo S, Park J, Lee J Y, Kweon I S. CBAM: Convolutional Block Attention Module. 2018. ECCV2018

Architecture

BAM

CBAM

Getting Started

$ git clone https://github.com/asdf2kr/BAM-CBAM-pytorch.git
$ cd BAM-CBAM-pytorch
$ python main.py --arch bam (default: bam network based on resnet50)

Performance

The table below shows models, dataset and performances

Model	Backbone	Dataset	Top-1	Top-5	Size
ResNet	resnet50	CIFAR-100	78.93%	-	23.70M
BAM	resnet50	CIFAR-100	79.62%	-	24.06M
CBAM	resnet50	CIFAR-100	81.02%	-	26.23M

Reference

Official PyTorch code