压缩激励机制驱动的尿液细胞图像分类算法  

Urine cell image classification algorithm based on the squeeze and excitation mechanism

作　　者：宋建锋[1] 韦玥 苗启广[1] 权义宁[1] 陈毓生 SONG Jianfeng;WEI Yue;MIAO Qiguang;QUAN Yining;CHEN Yusheng(School of Computer Science and Technology,Xidian University,Xi’an 710071,China;Unit 96963 of the People’s Liberation Army of China,Beijing 100000,China)

机构地区：[1]西安电子科技大学计算机科学与技术学院,陕西西安710071 [2]中国人民解放军96963部队,北京100000

出　　处：《西安电子科技大学学报》2020年第2期39-45,59,共8页Journal of Xidian University

基　　金：国家重点研发计划(*238*,2018YFC0807500);国家自然科学基金(61772396,61472302,61772392);西安市大数据与视觉智能关键技术重点实验室基金(201805053ZD4CG37);中央高校基本科研业务费专项资金(JB170304,JBF180301)。

摘　　要：针对尿液有形成分细胞图像的分类问题,提出了一种压缩激励机制驱动的GoogLeNet尿液细胞图像分类算法。该算法采用特征重标定机制,通过压缩操作和激励操作显式地建模Inception架构在训练过程中学习到的细胞特征通道之间的依赖关系,从而提升有用特征在当前任务中的重要程度。在尿液细胞数据集上的对比实验结果表明,在保证执行速度的情况下,该算法比GoogLeNet网络的分类准确率提升了3%,召回率提升了1%,具有更好的分类效果。In order to solve the classification problem on the urine-forming sub-cell images,a urine cell image classification algorithm based on the Squeeze-and-Excitation GoogLeNet is proposed.The algorithm uses the feature recalibration mechanism and brings about significant improvement in the useful feature for the current task through squeeze and excitation operations,which explicitly models interdependencies between cell feature channels learned by the Inception architecture during the training process.On the urine cell datasets,comparative experimental results show that the algorithm provides a better classification effect,which improves the accuracy of classification by 3% and the recall rate by 1%at the similar speed of the GoogLeNet network.

关 键 词：尿液细胞 压缩激励 特征重标定 图像分类 

分 类 号：TP37[自动化与计算机技术—计算机系统结构]