高技术通讯2024，Vol.34Issue(3)：233-247,15.DOI:10.3772/j.issn.1002-0470.2024.03.002
基于注意力与密集重参数化的目标检测算法
Object detection algorithm based on attention and dense reparameterization
摘要
Abstract
This paper presents an object detection algorithm using attention and dense reparameterization to tackle chal-lenges posed by complex backgrounds and variations in object sizes,which can adversely affect detection results.The proposed algorithm consists of two key components within an efficient feature extraction network based on CSP-DarkNet:the dense reparameterization module and the coordinate and spatial attention(CASA)module.The for-mer leverages dense connections to retain shallow features while reducing network complexity through reparameter-ization structures,while the CASA module captures necessary target information.Feature fusion is performed using feature pyramid network(FPN)and path aggregation network(PAN),and upsampling is achieved through content-aware reassembly of features(CARAFE),addressing the issue of insufficient capture of rich semantic information.To enhance model capabilities,a more efficient C3-G module is introduced to obtain gradient information,and depthwise separable convolution is employed to improve computational efficiency.Lastly,the detection output is enhanced by employing a cross-domain positive-negative sample matching strategy on a larger scale,augmenting positive samples and improving detection performance.Experimental results showcase the algorithm's advance-ments,achieving mAP@0.50 scores of 57.5%and 83.0%on the MS COCO and PASCAL VOC datasets,respec-tively.关键词
目标检测/重参数化/注意力机制/特征融合/上采样/正负样本匹配Key words
object detection/reparameterization/attention mechanism/feature fusion/upsampling/positive and negative sample matching引用本文复制引用
陈志旺,雷春明,吕昌昊,王婷,彭勇..基于注意力与密集重参数化的目标检测算法[J].高技术通讯,2024,34(3):233-247,15.基金项目
国家自然科学基金(61573305)和河北省自然科学基金(F2022203038,F2019203511)资助项目. （61573305）
