Lightweight Mixed-domain Attention Module

The official PyTorch implementation of "LMA: lightweight mixed-domain attention for efficient network design".

Method

Lightweight mixed-domain attention

Figure 1: An overview of the proposed LMA module.

LMA. Detail of implementations, including modules and the networks, can be found in Model in this repository.

Our environments and toolkits

• OS: Ubuntu 18.04.1
• CUDA: 11.6
• Python: 3.9.12
• Toolkit: PyTorch 1.10
• GPU: RTX A6000 (4x)
• ptflops
• For generating GradCAM++ results, please follow the code on this repository

Incorporate the LMA module into convolutional neural networks

Figure 2: The LMA-based inverted residual block.

Figure 3: The LMA-based residual block.

Overview of Results

Classification results on ImageNet-1K

Table 1: Comparison of different attention methods on the ImageNet validation set in terms of network parameters (# Params.), FLOPs, inference speed (frame per second, FPS), and Top-1/Top-5 accuracy (%).

Figure 4: Training and validation curves of LMA-based attention module under ResNet architectures with different depths.

Citation

If you find LMA useful in your research, please consider citing:

@article{yu2023lma,
	title={LMA: lightweight mixed-domain attention for efficient network design},
	author={Yu, Yang and Zhang, Yi and Song, Zhe and Tang, Cheng-Kai},
	journal={Applied Intelligence},
	volume={53},
	number={11},
	pages={13432--13451},
	year={2023},
	publisher={Springer}
}

Contact Information

If you have any suggestion or question, you can leave a message here or contact us directly: yang_y9802@163.com. Thanks for your attention!