Smart Agriculture ›› 2023, Vol. 5 ›› Issue (3): 96-109.doi: 10.12133/j.smartag.SA202308003
• Special Issue--Monitoring Technology of Crop Information • Previous Articles Next Articles
TANG Hui1(), WANG Ming2, YU Qiushi1, ZHANG Jiaxi1, LIU Liantao3, WANG Nan1()
Received:
2023-07-28
Online:
2023-09-30
Supported by:
TANG Hui, WANG Ming, YU Qiushi, ZHANG Jiaxi, LIU Liantao, WANG Nan. Root Image Segmentation Method Based on Improved UNet and Transfer Learning[J]. Smart Agriculture, 2023, 5(3): 96-109.
Add to citation manager EndNote|Ris|BibTeX
URL: http://www.smartag.net.cn/EN/10.12133/j.smartag.SA202308003
Table 2
Model improvement strategies and explanations for ablation experiments
模型改进策略 | 模型解释 |
---|---|
Conv_1+Concat | 进行一次完整卷积计算,并将其上采样和每层编码器中特征图拼接 |
Conv_2+Concat | 进行两次完整卷积计算,并将其上采样和每层编码器中特征图拼接 |
DP Conv+Concat | 进行两次深度可分离卷积计算,并将其上采样和每层编码器中特征图拼接 |
CBAM+Concat | 进行完整卷积后再进行注意力机制计算,并将其上采样和每层编码器中特征图拼接 |
Conv_1+Add | 进行一次完整卷积计算,并将其与跳跃连接的特征图相加并和上采样进行拼接 |
Conv_2+Add (本研究) | 进行两次完整卷积计算,并将其与跳跃连接的特征图相加并和上采样进行拼接 |
DP Conv+Add | 进行两次深度可分离卷积计算,并将其与跳跃连接的特征图相加并和上采样进行拼接 |
CBAM+Add | 进行完整卷积后,进行注意力机制计算,并将其与跳跃连接的特征图相加并和上采样进行拼接 |
Table 3
Evaluation Indicators for various improved models in ablation experiments
评估指标 | UNet | Conv_1+Concat | Conv_2+Concat | DP Conv+Concat | CBAM+Concat | Conv_1+Add | Conv_2+Add | DP Conv +Add | CBAM+Add |
---|---|---|---|---|---|---|---|---|---|
R IoU/% | 63.71 | 55.61 | 63.83 | 62.81 | 63.71 | 63.39 | 64.44 | 63.00 | 63.90 |
B IoU/% | 98.79 | 98.56 | 98.79 | 98.76 | 98.79 | 98.77 | 98.79 | 98.75 | 98.79 |
mIoU/% | 81.25 | 77.08 | 81.31 | 80.79 | 81.25 | 81.08 | 81.62 | 80.88 | 81.35 |
R Recall/% | 72.39 | 61.68 | 72.68 | 71.18 | 72.22 | 72.57 | 74.25 | 72.36 | 72.89 |
B Recall/% | 99.60 | 99.68 | 99.59 | 99.61 | 99.60 | 99.57 | 99.55 | 99.56 | 99.56 |
mRecall/% | 85.99 | 80.68 | 86.13 | 85.39 | 85.91 | 86.07 | 86.90 | 85.96 | 86.24 |
R Precision/% | 84.16 | 84.95 | 83.98 | 84.23 | 84.39 | 83.36 | 83.00 | 82.96 | 83.82 |
B Precision/% | 99.18 | 98.87 | 99.19 | 99.15 | 99.18 | 99.19 | 99.24 | 99.18 | 99.20 |
mPrecision/% | 91.67 | 91.91 | 91.59 | 91.69 | 91.78 | 91.27 | 91.12 | 91.07 | 91.51 |
R F 1/% | 77.83 | 71.47 | 77.92 | 77.16 | 77.83 | 77.59 | 78.38 | 77.30 | 77.97 |
B F 1/% | 99.39 | 99.27 | 99.39 | 99.38 | 99.39 | 99.38 | 99.39 | 99.37 | 99.38 |
Table 4
Evaluation indicators of each comparative model in comparative experiments
估计指标 | DeeplabV3Plus | PSPNet | SegNet | 改进模型(UNet+Conv_2+Add) |
---|---|---|---|---|
Root IoU/% | 64.00 | 54.33 | 63.08 | 64.44 |
Background IoU/% | 98.79 | 98.53 | 98.79 | 98.79 |
mIoU/% | 81.39 | 76.43 | 89.93 | 81.62 |
Root Recall/% | 73.53 | 59.51 | 73.86 | 74.25 |
Background Recall/% | 99.47 | 99.72 | 99.55 | 99.55 |
mRecall/% | 86.50 | 79.61 | 86.71 | 86.90 |
Root Precision/% | 81.18 | 86.17 | 82.87 | 83.00 |
Background Precision/% | 99.31 | 98.81 | 99.23 | 99.24 |
mPrecision/% | 90.24 | 92.49 | 91.05 | 91.12 |
Root F 1/% | 77.17 | 70.40 | 78.11 | 78.38 |
Background F 1/% | 99.39 | 99.26 | 99.39 | 99.39 |
Table 5
Four phenotypic data indicators for root phenotype determination
方法 | 总根长/px | 平均直径/px | 容量/px3 | 表面积/px2 |
---|---|---|---|---|
手工标注 | 281,884.9367 | 16.4984 | 86,505,316.2980 | 13,216,115.0220 |
改进模型UNet+Conv_2+Add | 236,648.6779 | 16.2529 | 90,592,259.8600 | 13,275,772.0500 |
PSPNet | 186,125.1123 | 14.1353 | 61,695,138.3499 | 9,377,353.0364 |
SegNet | 240,006.0245 | 15.7012 | 85,858,025.2235 | 12,975,598.3651 |
DeeplabV3Plus | 225,178.9484 | 15.6688 | 78,377,863.8983 | 12,039,045.5469 |
Table 6
Evaluation indicators of improved model(UNet+Conv_2+Add) and original model under ordinary training and transfer learning
评估指标 | UNet普通训练 | UNet 迁移学习 | 改进模型普通训练 | 改进模型迁移学习 |
---|---|---|---|---|
R IoU/% | 62.93 | 63.22 | 63.33 | 64.58 |
B IoU/% | 98.75 | 98.75 | 98.76 | 98.79 |
mIoU/% | 80.84 | 80.99 | 81.40 | 81.68 |
R Recall/% | 71.27 | 72.10 | 72.30 | 74.09 |
B Recall/% | 99.60 | 99.58 | 99.58 | 99.56 |
mRecall/% | 85.44 | 85.84 | 85.94 | 86.83 |
R Precision/% | 84.32 | 83.69 | 83.62 | 83.41 |
B Precision/% | 99.14 | 99.17 | 99.17 | 99.23 |
mPrecision/% | 91.73 | 91.43 | 91.40 | 91.32 |
R F 1/% | 77.25 | 77.46 | 77.55 | 78.47 |
B F 1/% | 99.37 | 99.37 | 99.37 | 99.39 |
1 |
王宁, 李继光, 娄翼来, 等. 作物根系形态对施肥措施的响应[J]. 中国农学通报, 2020, 36(3): 53-58.
|
|
|
2 |
|
3 |
吴茜, 张伟欣, 张玲玲, 等. 植物根系表型信息获取技术研究进展[J]. 江苏农业科学, 2021, 49(5): 31-37.
|
|
|
4 |
|
5 |
肖爽, 刘连涛, 张永江, 等. 植物微根系原位观测方法研究进展[J]. 植物营养与肥料学报, 2020, 26(2): 370-385.
|
|
|
6 |
赵先丽, 蔡福, 李荣平, 等. 春玉米根系图像语义分割最佳分辨率和概率阈值研究[J]. 核农学报, 2023, 37(8): 1690-1699.
|
|
|
7 |
何勇, 李禧尧, 杨国峰, 等. 室内高通量种质资源表型平台研究进展与展望[J]. 农业工程学报, 2022, 38(17): 127-141.
|
|
|
8 |
|
9 |
|
10 |
|
11 |
|
12 |
|
13 |
DAS A,
|
14 |
|
15 |
|
16 |
|
17 |
|
18 |
|
19 |
|
20 |
|
21 |
|
22 |
|
23 |
|
24 |
|
25 |
|
26 |
|
27 |
|
28 |
林娜, 何静, 王斌, 等. 结合植被光谱特征与Sep-UNet的城市植被信息智能提取方法[J]. 地球信息科学学报, 2023, 25(8): 1717-1729.
|
|
|
29 |
申传庆, 王凯, 王文杰. 基于ResNet-UNet的地表覆盖自动分类技术研究[J]. 地理空间信息, 2023, 21(6): 21-23, 27.
|
|
|
30 |
陈桂芬, 赵姗, 曹丽英, 等. 基于迁移学习与卷积神经网络的玉米植株病害识别[J]. 智慧农业, 2019, 1(2): 34-44.
|
|
|
31 |
|
32 |
|
[1] | PAN Weiting, SUN Mengli, YUN Yan, LIU Ping. Identification Method of Wheat Grain Phenotype Based on Deep Learning of ImCascade R-CNN [J]. Smart Agriculture, 2023, 5(3): 110-120. |
[2] | GUAN Bolun, ZHANG Liping, ZHU Jingbo, LI Runmei, KONG Juanjuan, WANG Yan, DONG Wei. The Key Issues and Evaluation Methods for Constructing Agricultural Pest and Disease Image Datasets: A Review [J]. Smart Agriculture, 2023, 5(3): 17-34. |
[3] | LONG Jianing, ZHANG Zhao, LIU Xiaohang, LI Yunxia, RUI Zhaoyu, YU Jiangfan, ZHANG Man, FLORES Paulo, HAN Zhexiong, HU Can, WANG Xufeng. Wheat Lodging Types Detection Based on UAV Image Using Improved EfficientNetV2 [J]. Smart Agriculture, 2023, 5(3): 62-74. |
[4] | ZHANG Gan, YAN Haifeng, HU Gensheng, ZHANG Dongyan, CHENG Tao, PAN Zhenggao, XU Haifeng, SHEN Shuhao, ZHU Keyu. Identification Method of Wheat Field Lodging Area Based on Deep Learning Semantic Segmentation and Transfer Learning [J]. Smart Agriculture, 2023, 5(3): 75-85. |
[5] | LIU Yixue, SONG Yuyang, CUI Ping, FANG Yulin, SU Baofeng. Diagnosis of Grapevine Leafroll Disease Severity Infection via UAV Remote Sensing and Deep Learning [J]. Smart Agriculture, 2023, 5(3): 49-61. |
[6] | MAO Kebiao, ZHANG Chenyang, SHI Jiancheng, WANG Xuming, GUO Zhonghua, LI Chunshu, DONG Lixin, WU Menxin, SUN Ruijing, WU Shengli, JI Dabin, JIANG Lingmei, ZHAO Tianjie, QIU Yubao, DU Yongming, XU Tongren. The Paradigm Theory and Judgment Conditions of Geophysical Parameter Retrieval Based on Artificial Intelligence [J]. Smart Agriculture, 2023, 5(2): 161-171. |
[7] | PAN Chenlu, ZHANG Zhenghua, GUI Wenhao, MA Jiajun, YAN Chenxi, ZHANG Xiaomin. Rice Disease and Pest Recognition Method Integrating ECA Mechanism and DenseNet201 [J]. Smart Agriculture, 2023, 5(2): 45-55. |
[8] | ZHU Haipeng, ZHANG Yu'an, LI Huanhuan, WANG Jianwen, YANG Yingkui, SONG Rende. Classification and Recognition Method for Yak Meat Parts Based on Improved Residual Network Model [J]. Smart Agriculture, 2023, 5(2): 115-125. |
[9] | XIA Xue, CHAI Xiujuan, ZHANG Ning, ZHOU Shuo, SUN Qixin, SUN Tan. A Lightweight Fruit Load Estimation Model for Edge Computing Equipment [J]. Smart Agriculture, 2023, 5(2): 1-12. |
[10] | WANG Yapeng, CAO Shanshan, LI Quansheng, SUN Wei. Desert Plant Recognition Method Under Natural Background Incorporating Transfer Learning and Ensemble Learning [J]. Smart Agriculture, 2023, 5(2): 93-103. |
[11] | HU Songtao, ZHAI Ruifang, WANG Yinghua, LIU Zhi, ZHU Jianzhong, REN He, YANG Wanneng, SONG Peng. Extraction of Potato Plant Phenotypic Parameters Based on Multi-Source Data [J]. Smart Agriculture, 2023, 5(1): 132-145. |
[12] | GUO Yangyang, DU Shuzeng, QIAO Yongliang, LIANG Dong. Advances in the Applications of Deep Learning Technology for Livestock Smart Farming [J]. Smart Agriculture, 2023, 5(1): 52-65. |
[13] | ZUO Min, HU Tianyu, DONG Wei, ZHANG Kexin, ZHANG Qingchuan. Forecast and Analysis of Agricultural Products Logistics Demand Based on Informer Neural Network: Take the Central China Aera as An Example [J]. Smart Agriculture, 2023, 5(1): 34-43. |
[14] | ZHANG Wenjing, JIANG Zezhong, QIN Lifeng. Identifying Multiple Apple Leaf Diseases Based on the Improved CBAM-ResNet18 Model Under Weak Supervision [J]. Smart Agriculture, 2023, 5(1): 111-121. |
[15] | JI Jie, JIN Zhou, WANG Rujing, LIU Haiyan, LI Zhiyuan. Progressive Convolutional Net Based Method for Agricultural Named Entity Recognition [J]. Smart Agriculture, 2023, 5(1): 122-131. |
Viewed | ||||||
Full text |
|
|||||
Abstract |
|
|||||