Spectroscopy and Spectral Analysis, Volume. 41, Issue 7, 2171(2021)
Hyperspectral Imaging for Detection of Leguminivora Glycinivorella Based on 3D Few-Shot Meta-Learning Model
Figures & Tables(4)
Fig. 3. Hyperspectral images of soybean samples
(a): Normal soybean; (b): Soybean with egg;(c): Soybean with larvae; (d): Gnawed soybean
Table 1. Detection results of different classification models in the cases of 4-way 1-shot and 4-way 5-shot
View tableView in ArticleTable 1. Detection results of different classification models in the cases of 4-way 1-shot and 4-way 5-shot
小样本
学习模型特征提取
模型学习率 正确率/% 1-shot 5-shot MAML ResNet18 0.01 50.98±1.19 59.94±0.56 0.001 43.43±0.55 48.03±0.23 Conv4 0.01 42.15±0.38 52.65±0.79 0.001 35.15±0.73 37.06±0.58 MM ResNet18 0.01 69.19±1.21 78.65±0.79 0.001 56.58±0.47 69.49±0.82 Conv4 0.01 52.46±0.40 64.09±0.70 0.001 48.07±0.97 60.21±1.15 3D-RN ResNet18 0.01 71.12±0.61 82.75±2.50 0.001 65.90±0.77 76.40±1.23 Conv4 0.01 57.31±0.98 69.12±0.31 0.001 53.33±0.96 65.15±1.17
Tools
Get Citation
Copy Citation Text
Jiang-sheng GUI, Jing-yi FEI, Xia-ping FU. Hyperspectral Imaging for Detection of Leguminivora Glycinivorella Based on 3D Few-Shot Meta-Learning Model[J]. Spectroscopy and Spectral Analysis, 2021, 41(7): 2171
Paper Information
Category: Research Articles
Received: Jul. 11, 2020
Accepted: --
Published Online: Sep. 8, 2021
The Author Email: GUI Jiang-sheng (jsgui@zstu.edu.cn)
© Copyright 2018-2021 | Chinese Laser Press.
All Rights Reserved 沪ICP备15018463号-20