Research on near-ground forage hyperspectral imagery classification based on fusion preprocessing process
文献类型: 外文期刊
作者: Yilei Liu;Xin Pan;Jiangping Liu;Shengwei Zhang;Weihong Yan
作者机构:
关键词: attention mechanism;CBAMResNet;classification;Forage;HSI
期刊名称: International Journal of Digital Earth
ISSN: 1753-8947
年卷期: 2023 年 16 卷 2 期
页码:
收录情况: SCIE(2023版) ; ; EI(2023版)
摘要: Accurate identification and classification of forage grass are pivotal in optimizing forage resources and breeding superior forage varieties. Given the low accuracy in forage image identification and classification, and the loss of some features from preprocessing, we proposed an innovative approach that integrates preprocessing operations directly into the model instead of preceding feature analysis. We captured near-ground hyperspectral imagery of forage in the field and applied two deep learning models–Squeeze and Excitation ResNet (SEResNet) and Convolution Block Attention Module ResNet (CBAMResNet). These models not only harness the automatic learning capabilities of the ResNet deep network but also employ channel attention and a channel-plus-space dual attention mechanism to filter and label important features. This approach enhances data extraction and analysis, strengthen the correlation between the channel and space dimensions while eliminating redundancy and noise. We compared the performance of the proposed methods with the current popular methods by six evaluation parameters, including overall accuracy (OA), average accuracy (AA), Kappa coefficient, etc. Experiment results show the OA of SEResNet and CBAMResNet are 96.57% and 98.35% respectively. The experiments demonstrate the feasibility of incorporating preprocessing into the network and the effectiveness of the new idea for the classification research of forage.
分类号:
- 相关文献
作者其他论文 更多>>
-
Rapid on-site differentiation of two invasive whitefly cryptic species using LAMP
作者:Yage Liu;Lifeng Yang;Yilei Liu;Yu Chen;Youjun Zhang;Dong Chu
关键词:Cryptic species;Loop-mediated isothermal amplification;Low-temperature preservation;On-site diagnostics;Reaction mixture
-
Semantic segmentation of remote sensing images based on multiscale features and global information modeling
作者:Yupeng Gao;Xiaoling Luo;Xiaojing Gao;Weihong Yan;Xin Pan;Xueliang Fu
关键词:Global modeling information;Multiscale features;Remote sensing image;Semantic segmentation
-
Effective strategy for alleviation of oil contamination in marine via the superhydrophobic biochar: Performances, mechanisms and environmental recalcitrance
作者:Xin Pan;Shaoping Kuang;Xiao Wang
关键词:Environmental recalcitrance;Layered double hydroxides;Oil spill remediation;Superhydrophobic wettability
-
TMNet: A Two-Branch Multi-Scale Semantic Segmentation Network for Remote Sensing Images
作者:Yupeng Gao;Shengwei Zhang;Dongshi Zuo;Weihong Yan;Xin Pan
关键词:global modeling;remote sensing images;semantic segmentation;Swin transformer
-
Multi-Temporal Hyperspectral Classification of Grassland Using Transformer Network
作者:Xuanhe Zhao;Shengwei Zhang;Ruifeng Shi;Weihong Yan;Xin Pan
关键词:grassland;hyperspectral classification;multi-temporal;transformer network
-
CO2 and H2O as sweep gases elevated carbon stability and decreased phytotoxicity of biochars
作者:Xin Pan;Xiao Wang;Liuqingqing Liu;Shaoping Kuang;Hao Zheng
关键词:Carbon sequestration;Ecological risk;Phytotoxicity;Stability;Sweep gas
-
Research Progress of the Improved Saline-Alkaline Soil by Biochar
作者:Wenzhu Yang;Hong Nan;Weihong Yan;Yan Jiao
关键词: