基于深度学习的语义分割网络  

Semantic Segmentation Network Based on Deep Learning

作　　者：代具亭 汤心溢[1,3] 刘鹏[1,3] DAI Ju-ting;TANG Xin-yi;LIU Peng(Shanghai Institute of Technical Physics,Chinese Academy of Sciences,Shanghai 200083,China;University of Chinese Academy of Sciences,Beijing 100084,China;Key Laboratory of Infrared System Detection and Imaging Technology, Chinese Academy of Sciences,Shanghai 200083,China)

机构地区：[1]中国科学院上海技术物理研究所,上海200083 [2]中国科学院大学,北京100084 [3]中国科学院红外探测与成像技术重点实验室,上海200083

出　　处：《红外》2018年第4期33-38,48,共7页Infrared

基　　金：国家"十三五"国防预研项目(Jzx2016-0404/Y72-2);中国科学院青年创新促进会项目(2014216);上海市现场物证重点实验室基金项目(2017xcwzk08)

摘　　要：提出了一种基于深度学习的语义分割网络。该网络通过多孔卷积设计了一个能提取图像多尺度信息的空间金字塔模块,并通过大量实验探索了空间金字塔模块中多孔采样率和多尺度分支对于网络场景解析能力的影响。讨论了网络训练中不同超参数对于网络性能的影响。在SUN RGB-D数据集上的测试结果显示,与其它state-of-the-art的语义分割网络相比,本文设计的网络性能突出。最后,还对基于红外图像的语义分割进行了初步探索。A semantic segmentation network based on deep learning is proposed.The network designs a spatial pyramid module which can extract multi-scale information from images through Atrous convolution.It also explores the influence of Atrous convolution sampling rate and multi-scale branches on the performance of network through extensive experiment.the impact of hyperparameters on network performance during training is discussed.The test results on the SUN RGB-D dataset show that compared with other state-of-the-art semantic segmentation networks,the performance of the network we proposed is outstanding.Finally,the semantic segmentation based on infrared images is explored preliminarily.

关 键 词：卷积神经网络 语义分割 多尺度 红外图像 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]