Real-Time Dense Semantic Labeling with Dual-Path Framework for High-Resolution Remote Sensing Image

被引:14
|
作者
Wang, Yuhao [1 ,2 ]
Chen, Chen [3 ]
Ding, Meng [4 ]
Li, Jiangyun [1 ,2 ]
机构
[1] Univ Sci & Technol Beijing, Sch Automat & Elect Engn, Beijing 100083, Peoples R China
[2] Minist Educ, Key Lab Knowledge Automat Ind Proc, Beijing 100083, Peoples R China
[3] Univ N Carolina, Dept Elect & Comp Engn, Charlotte, NC 28223 USA
[4] Thermo Fisher Sci, Richardson, TX 75081 USA
基金
中国国家自然科学基金;
关键词
remote sensing image; real-time dense semantic labeling; convolutional neural networks; light-weight;
D O I
10.3390/rs11243020
中图分类号
X [环境科学、安全科学];
学科分类号
08 ; 0830 ;
摘要
Dense semantic labeling plays a pivotal role in high-resolution remote sensing image research. It provides pixel-level classification which is crucial in land cover mapping and urban planning. With the recent success of the convolutional neural network (CNN), accuracy has been greatly improved by previous works. However, most networks boost performance by involving too many parameters and computational overheads, which results in more inference time and hardware resources, while some attempts with light-weight networks do not achieve satisfactory results due to the insufficient feature extraction ability. In this work, we propose an efficient light-weight CNN based on dual-path architecture to address this issue. Our model utilizes three convolution layers as the spatial path to enhance the extraction of spatial information. Meanwhile, we develop the context path with the multi-fiber network (MFNet) followed by the pyramid pooling module (PPM) to obtain a sufficient receptive field. On top of these two paths, we adopt the channel attention block to refine the features from the context path and apply a feature fusion module to combine spatial information with context information. Moreover, a weighted cascade loss function is employed to enhance the learning procedure. With all these components, the performance can be significantly improved. Experiments on the Potsdam and Vaihingen datasets demonstrate that our network performs better than other light-weight networks, even some classic networks. Compared to the state-of-the-art U-Net, our model achieves higher accuracy on the two datasets with 2.5 times less network parameters and 22 times less computational floating point operations (FLOPs).
引用
收藏
页数:20
相关论文
共 50 条
  • [31] Deep Semantic Understanding of High Resolution Remote Sensing Image
    Qu, Bo
    Li, Xuelong
    Tao, Dacheng
    Lu, Xiaoqiang
    2016 INTERNATIONAL CONFERENCE ON COMPUTER, INFORMATION AND TELECOMMUNICATION SYSTEMS (CITS), 2016, : 124 - 128
  • [32] Real-Time Moving Object Detection in High-Resolution Video Sensing
    Zhu, Haidi
    Wei, Haoran
    Li, Baoqing
    Yuan, Xiaobing
    Kehtarnavaz, Nasser
    SENSORS, 2020, 20 (12) : 1 - 15
  • [33] RAANet: A Residual ASPP with Attention Framework for Semantic Segmentation of High-Resolution Remote Sensing Images
    Liu, Runrui
    Tao, Fei
    Liu, Xintao
    Na, Jiaming
    Leng, Hongjun
    Wu, Junjie
    Zhou, Tong
    REMOTE SENSING, 2022, 14 (13)
  • [34] Dual-Path Adversarial Generation Network for Super-Resolution Reconstruction of Remote Sensing Images
    Ren, Zhipeng
    Zhao, Jianping
    Chen, Chunyi
    Lou, Yan
    Ma, Xiaocong
    APPLIED SCIENCES-BASEL, 2023, 13 (03):
  • [35] A decoupled search deep network framework for high-resolution remote sensing image classification
    Wang, Kun
    Han, Ling
    Li, Liangzhi
    REMOTE SENSING LETTERS, 2023, 14 (03) : 243 - 253
  • [36] Intelligent remote sensing satellite and remote sensing image real-time service
    Wang M.
    Yang F.
    Cehui Xuebao/Acta Geodaetica et Cartographica Sinica, 2019, 48 (12): : 1586 - 1594
  • [37] ORBNet: Original Reinforcement Bilateral Network for High-Resolution Remote Sensing Image Semantic Segmentation
    Zhang, Yijie
    Cheng, Jian
    Su, Yanzhou
    Wu, Yuheng
    Ma, Qijun
    IEEE JOURNAL OF SELECTED TOPICS IN APPLIED EARTH OBSERVATIONS AND REMOTE SENSING, 2024, 17 : 15900 - 15913
  • [38] High-resolution remote sensing image semantic segmentation based on a deep feature aggregation network
    Wang, Zhen
    Guo, Jianxin
    Huang, Wenzhun
    Zhang, Shanwen
    MEASUREMENT SCIENCE AND TECHNOLOGY, 2021, 32 (09)
  • [39] DNAS: Decoupling Neural Architecture Search for High-Resolution Remote Sensing Image Semantic Segmentation
    Wang, Yu
    Li, Yansheng
    Chen, Wei
    Li, Yunzhou
    Dang, Bo
    REMOTE SENSING, 2022, 14 (16)
  • [40] Ulva polifera Detection from High Resolution Remote Sensing Images Based on Dual-Path Convolutional Neural Networks
    Wang, Yanli
    Dong, Zhipeng
    Wang, Mi
    Wuhan Daxue Xuebao (Xinxi Kexue Ban)/Geomatics and Information Science of Wuhan University, 49 (12): : 2261 - 2270