Bilateral U-Net semantic segmentation with spatial attention mechanism

作     者：Guangzhe Zhao Yimeng Zhang Maoning Ge Min Yu 

作者机构：Beijing University of Civil Engineering and ArchitectureCollege of Electrical and Information EngineeringBeijingChina Graduate School of InformaticsNagoya UniversityNagoyaJapan Department of General SurgeryGuangdong Provincial People's HospitalGuangdong Academy of Medical SciencesGuangzhouChina 

出 版 物：《CAAI Transactions on Intelligence Technology》 (智能技术学报（英文）)

年 卷 期：2023年第8卷第2期

页      面：297-307页

核心收录：

学科分类：04[教育学] 0812[工学-计算机科学与技术（可授工学、理学学位）] 

基　　金：Ministry of Science and Technology Basic Resources Survey Special Project,Grant/Award Number:2019FY100900 High-level Hospital Construction Project,Grant/Award Number:DFJH2019015 National Natural Science Foundation of China,Grant/Award Number:61871021 Guangdong Natural Science Foundation,Grant/Award Number:2019A1515011676 Beijing Key Laboratory of Robotics Bionic and Functional Research。 

主　　题：attention mechanism receptive field semantic fusion semantic segmentation spatial attention module U-Net 

摘      要：Aiming at the problem that the existing models have a poor segmentation effect on imbalanced data sets with small-scale samples,a bilateral U-Net network model with a spatial attention mechanism is designed.The model uses the lightweight MobileNetV2 as the backbone network for feature hierarchical extraction and proposes an Attentive Pyramid Spatial Attention(APSA)module compared to the Attenuated Spatial Pyramid module,which can increase the receptive field and enhance the information,and finally adds the context fusion prediction branch that fuses high-semantic and low-semantic prediction results,and the model effectively improves the segmentation accuracy of small data sets.The experimental results on the CamVid data set show that compared with some existing semantic segmentation networks,the algorithm has a better segmentation effect and segmentation accuracy,and its mIOU reaches 75.85%.Moreover,to verify the generality of the model and the effectiveness of the APSA module,experiments were conducted on the VOC 2012 data set,and the APSA module improved mIOU by about 12.2%.