Research on Extracting Special Plant Planting Plots from High-resolution Remote Sensing Images Using I-PSPNet Semantic Segmentation Model

超参数	Value
优化方法	Adam
初始学习率	0.000 5
学习率调整	每过一个轮次学习率变为原来的0.94
批大小（Batch Size）	8
训练轮数（Training Epochs）	200
早停（Early Stopping）	训练损失10个轮次不下降

Table 2. Performance comparison of PSPNet models before and after backbone network improvement
View table
View in Article
Table 2. Performance comparison of PSPNet models before and after backbone network improvement
模型骨干网络测试速度¹(fps) MPA(%) MIoU(%)
PSPNet ResNet50 44.4 87.30 79.73
PSPNet MobileNetv2 96.0 86.84 78.32

Table 3. Performance comparison of PSPNet model before and after adding SE module
View table
View in Article
Table 3. Performance comparison of PSPNet model before and after adding SE module
模型 MPA(%) MIoU(%)
PSPNet 87.30 79.73
PSPNet+SE 92.60 84.57

Table 4. Performance comparison of PSPNet model before and after adding encoder-decoder structure
View table
View in Article
Table 4. Performance comparison of PSPNet model before and after adding encoder-decoder structure
模型 MPA/% MIoU/%
PSPNet 87.30 79.73
PSPNet+Encoder-Decoder 88.36 80.70

Table 5. Performance comparison of PSPNet models before and after loss function improvement
View table
View in Article
Table 5. Performance comparison of PSPNet models before and after loss function improvement
模型 Loss MPA/% MIoU/%
PSPNet CE Loss 87.30 79.73
PSPNet CE Loss+Dice loss 88.96 80.21

Table 6. Comparison of Extraction Performance of Poppy Plots with Semantic Segmentation Models
View table
View in Article
Table 6. Comparison of Extraction Performance of Poppy Plots with Semantic Segmentation Models
模型测试速度¹/fps MPA/% MIoU/%
UNet 26.9 71.00 65.65
Deeplabv3+ 28.1 87.59 80.25
PSPNet 44.4 87.30 79.73
I-PSPNet 84.2 95.00 84.59

Table 7. Comparison of test results between NGB band dataset and RGB band dataset
View table
View in Article
Table 7. Comparison of test results between NGB band dataset and RGB band dataset
数据集波段 MPA/% MIoU/%
NGB 95.00 84.59
RGB 92.28 83.42

Tools

Get Citation

Copy Citation Text

Zhigang LU, Fangmiao CHEN, Chao YUAN, Yichen TIAN, Qiang CHEN, Meiping WEN, Kai YIN, Guang YANG. Research on Extracting Special Plant Planting Plots from High-resolution Remote Sensing Images Using I-PSPNet Semantic Segmentation Model[J]. Remote Sensing Technology and Application, 2024, 39(1): 222

Download Citation

EndNote(RIS)BibTex Plain Text

Set citation alerts for article

Save article for my favorites

Paper Information

Category: Research Articles

Received: Mar. 26, 2022

Accepted: --

Published Online: Jul. 22, 2024

The Author Email: Zhigang LU (2233758751@qq.com)

DOI:10.11873/j.issn.1004-0323.2024.1.0222

Topics

laser devices and laser physics

Lasers and Laser Optics

Laser physics

laser manufacturing

Instrumentation, Measurement and Metrology

Table 1. Super parameters settings of semantic segmentation models

Table 1. Super parameters settings of semantic segmentation models

Table 2. Performance comparison of PSPNet models before and after backbone network improvement

Table 2. Performance comparison of PSPNet models before and after backbone network improvement

Table 3. Performance comparison of PSPNet model before and after adding SE module

Table 3. Performance comparison of PSPNet model before and after adding SE module

Table 4. Performance comparison of PSPNet model before and after adding encoder-decoder structure

Table 4. Performance comparison of PSPNet model before and after adding encoder-decoder structure

Table 5. Performance comparison of PSPNet models before and after loss function improvement

Table 5. Performance comparison of PSPNet models before and after loss function improvement

Table 6. Comparison of Extraction Performance of Poppy Plots with Semantic Segmentation Models

Table 6. Comparison of Extraction Performance of Poppy Plots with Semantic Segmentation Models

Table 7. Comparison of test results between NGB band dataset and RGB band dataset

Table 7. Comparison of test results between NGB band dataset and RGB band dataset