高技术通讯2017，Vol.27Issue(9)：808-815,8.DOI:10.3772/j.issn.1002-0470.2017.09-10.005
图像语义分割深度学习模型综述
Survey of the deep learning models for image semantic segmentation
摘要
Abstract
The concept that image semantic segmentation is essentially the dense classification in pixel level , as well as its core position and practical significance are interpreted .Then, the commonly used classifications and latest achievements in image semantic segmentation are comprehensively reviewed , and for some deep learning models for image semantic segmentation , the pixel accuracy , average pixel accuracy , mean intersection over union and fre-quency-weighted intersection over union on the PASCAL VOC 2012 dataset are compared in detail .Meanwhile , the models' other performance indexes including the average time-consuming, program framework, language used, code readability , difficulty of deployment are shown .Finally, the developments of image semantic segmentation are summarized and discussed , and the challeges facing the models such as lack of training data sets , difficulty of pa-rameter optimization and single structure are pointed out .关键词
深度学习/语义分割/PASCAL VOC/卷积神经网络Key words
deep learning/semantic segmentation/PASCAL VOC/convolutional neural networks引用本文复制引用
张新明,祝晓斌,蔡强,刘新亮,邵玮,王磊..图像语义分割深度学习模型综述[J].高技术通讯,2017,27(9):808-815,8.基金项目
国家自然科学基金(61402023),北京市自然科学基金(4162019)和食品安全知识图谱及大数据平台研制(Z161100001616004)资助项目. （61402023）
