A Dual-branch Fusion Attention Mechanism for Lightweight Networks
DENG Yuhan
YANG Fumin
YUAN Ling
HU Guanrong
Abstract：The attention mechanism attracts more and more attention.Many studies respectively proved the effectiveness of channel attention and spatial attention in improving model performance.However,existing algorithms usually ignore how to well combine these two kinds of information.In this regard,by effectively combining channel attention,spatial attention,and feature in-formation extracted globally,a new attention mechanism for mobile networks called dual-branch fusion attention is proposed,and is applied to several classic lightweight networks for experimentation.The experimental results show that the accuracy of the model in-troduced with the dual-branch fusion attention mechanism is obviously higher than that of the original model on the CIFAR-100 and ImageNet-100 datasets,and the amount of floating-point calculations and the model volume do not increase significantly.
Keywords：lightweight networkattention mechanismdouble branch fusion
Publication Date：2023-12-20
Online Publishing Date：2025-08-15(First online date of this platform, not the publication date of the document)
Pages：6( 2831-2835,3027 )