跳到主要导航 跳到搜索 跳到主要内容

Spatial-Spectral Contrastive Self-Supervised Learning With Dual Path Networks for Hyperspectral Target Detection

  • School of Geophysics and Geomatics

科研成果: 期刊稿件 › 论文 › 同行评审

35 引用 (Scopus)

摘要

Deep learning (DL) has shown great success in hyperspectral target detection (HTD). Because of the limited targets of interest in the hyperspectral image (HSI), many DL-based methods focus on spectral feature extraction by expanding samples, exhibiting low model transferability, as well as high time consumption. To address these issues, we propose a novel spatial-spectral contrastive self-supervised learning-based HTD method with dual path networks (DPN-CSSTD). This method constructs a contrastive task using an HSI augmentation method and a DPN encoder on a large amount of unlabeled data, enabling the pretrained model with spatial-spectral discrimination ability. The DPN encoder uses the convolution neural network and skip connections to improve the spatial-spectral feature extraction. Additionally, the proposed method constructs a transfer learning model with the pretrained DPN encoder and a newly added detector to improve the pretrained model transferability and the time efficiency of target detection (TD) tasks. The transfer learning model is fine-tuned via a proposed fine-tuning strategy with a few selected samples to adapt to different TD tasks. Extensive experiments conducted on four challenging hyperspectral datasets demonstrate that the proposed method outperforms other state-of-the-art approaches.
源语言英语
期刊论文编号5515612
页(从-至)1-1
页数12
期刊IEEE Transactions on Geoscience and Remote Sensing
卷62
DOI
出版状态已出版 - 1月 2024

基金资助

No Statement Available

关键词

  • Contrastive learning
  • convolution neural network (CNN)
  • hyperspectral target detection (HTD)
  • self-supervised learning (SSL)
  • simple linear iterative clustering (SLIC) segmentation

收录

  • Scopus
  • EI
  • SCIE

学术指纹

探究 'Spatial-Spectral Contrastive Self-Supervised Learning With Dual Path Networks for Hyperspectral Target Detection' 的科研主题。它们共同构成独一无二的学术指纹。

引用此