基于特征融合和注意力机制的SSD改进算法  

Improved SSD Algorithm based on Feature Fusion and Attention Mechanism

作　　者：秦大勇 林玉娥[1] 梁兴柱[1,2] QIN Dayong;LIN Yue;LIANG Xingzhu(School of computer science and Engineering,Anhui University of Science and Technology,Huainan Anhui 232001,China;Institute of Environment-friendly Materials and Occupational Health of Anhui University of Science and Technology(Wuhu),Wuhu Anhui 241003,China)

机构地区：[1]安徽理工大学计算机科学与工程学院,安徽淮南232001 [2]安徽理工大学环境友好材料与职业健康研究院(芜湖),安徽芜湖241003

出　　处：《兰州工业学院学报》2022年第6期72-77,共6页Journal of Lanzhou Institute of Technology

基　　金：芜湖市科技计划项目(2020yf48);安徽理工大学环境友好材料与职业健康研究院开发专项基金(ALW2021YF04);安徽理工大学研究生创新基金(2021CX2105)。

摘　　要：SSD算法在应用于目标检测任务时,对小目标的检测准确率较低,针对此问题本文提出了一种新的改进的SSD目标检测算法。该算法采用了ResNet50网络作为特征提取网络,加入了特征融合模块和注意力模块用以提升模型的检测能力。通过特征融合模块得到比原来单一特征层更具有判别能力的新特征层,更加有效的提取图像中的细节特征;经注意力模块对特征图中的通道进行加权处理,使模型更容易关注到有效特征层中的重要信息,抑制对背景等无用信息的关注度。实验表明:该算法在PASCAL VOC数据集上的mAP为81.2%,检测能力相较于传统的SSD算法得到了显著提升。An improved SSD objection detection algorithm is proposed in this paper to address the low detection accuracy of small targets when applied to objection detection tasks.The ResNet50 network is used in the algorithm as the feature extraction network and a feature fusion module and an attention module are added to improve the detection capability of the model.Through the feature fusion module,a new feature layer with more discriminative ability than that with the original single feature layer is obtained to extract the details of the image more effectively.Through the attention module,the channels in the feature map are weighted to make the model focus more easily on the important information in the effective feature layer and suppress the attention to the useless information such as background.The experiments show that the algorithm has a mAP of 81.2%on the PASCAL VOC dataset,and the detection capability is improved compared to the traditional SSD algorithm.

关 键 词：目标检测 特征提取网络 特征融合 注意力机制 SSD 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]