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
Foundation items:
About author:
TANG Hui, E-mail:318901249@qq.com
corresponding author:
WANG Nan, E-mail:cmwn@163.com
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: https://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] | LUO Youlu, PAN Yonghao, XIA Shunxing, TAO Youzhi. Lightweight Apple Leaf Disease Detection Algorithm Based on Improved YOLOv8 [J]. Smart Agriculture, 2024, 6(5): 128-138. |
[2] | LIU Yi, ZHANG Yanjun. ReluformerN: Lightweight High-Low Frequency Enhanced for Hyperspectral Agricultural Lancover Classification [J]. Smart Agriculture, 2024, 6(5): 74-87. |
[3] | NIAN Yue, ZHAO Kaixuan, JI Jiangtao. Cow Hoof Slippage Detecting Method Based on Enhanced DeepLabCut Model [J]. Smart Agriculture, 2024, 6(5): 153-163. |
[4] | ZHANG Yanqi, ZHOU Shuo, ZHANG Ning, CHAI Xiujuan, SUN Tan. A Regional Farming Pig Counting System Based on Improved Instance Segmentation Algorithm [J]. Smart Agriculture, 2024, 6(4): 53-63. |
[5] | WENG Zhi, FAN Qi, ZHENG Zhiqiang. Automatic Measurement Method of Beef Cattle Body Size Based on Multimodal Image Information and Improved Instance Segmentation Network [J]. Smart Agriculture, 2024, 6(4): 64-75. |
[6] | HOU Yiting, RAO Yuan, SONG He, NIE Zhenjun, WANG Tan, HE Haoxu. A Rapid Detection Method for Wheat Seedling Leaf Number in Complex Field Scenarios Based on Improved YOLOv8 [J]. Smart Agriculture, 2024, 6(4): 128-137. |
[7] | LI Hao, DU Yuqiu, XIAO Xingzhu, CHEN Yanxi. Remote Sensing Identification Method of Cultivated Land at Hill County of Sichuan Basin Based on Deep Learning [J]. Smart Agriculture, 2024, 6(3): 34-45. |
[8] | NIE Ganggang, RAO Honghui, LI Zefeng, LIU Muhua. Severity Grading Model for Camellia Oleifera Anthracnose Infection Based on Improved YOLACT [J]. Smart Agriculture, 2024, 6(3): 138-147. |
[9] | ZHANG Jing, ZHAO Zexuan, ZHAO Yanru, BU Hongchao, WU Xingyu. Oilseed Rape Sclerotinia in Hyperspectral Images Segmentation Method Based on Bi-GRU and Spatial-Spectral Information Fusion [J]. Smart Agriculture, 2024, 6(2): 40-48. |
[10] | PANG Chunhui, CHEN Peng, XIA Yi, ZHANG Jun, WANG Bing, ZOU Yan, CHEN Tianjiao, KANG Chenrui, LIANG Dong. HI-FPN: A Hierarchical Interactive Feature Pyramid Network for Accurate Wheat Lodging Localization Across Multiple Growth Periods [J]. Smart Agriculture, 2024, 6(2): 128-139. |
[11] | ZHANG Yuyu, BING Shuying, JI Yuanhao, YAN Beibei, XU Jinpu. Grading Method of Fresh Cut Rose Flowers Based on Improved YOLOv8s [J]. Smart Agriculture, 2024, 6(2): 118-127. |
[12] | ZHANG Jianhua, YAO Qiong, ZHOU Guomin, WU Wendi, XIU Xiaojie, WANG Jian. Intelligent Identification of Crop Agronomic Traits and Morphological Structure Phenotypes: A Review [J]. Smart Agriculture, 2024, 6(2): 14-27. |
[13] | GUO Wang, YANG Yusen, WU Huarui, ZHU Huaji, MIAO Yisheng, GU Jingqiu. Big Models in Agriculture: Key Technologies, Application and Future Directions [J]. Smart Agriculture, 2024, 6(2): 1-13. |
[14] | WANG Herong, CHEN Yingyi, CHAI Yingqian, XU Ling, YU Huihui. Image Segmentation Method Combined with VoVNetv2 and Shuffle Attention Mechanism for Fish Feeding in Aquaculture [J]. Smart Agriculture, 2023, 5(4): 137-149. |
[15] | LI Zhengkai, YU Jiahui, PAN Shijia, JIA Zefeng, NIU Zijie. Individual Tree Skeleton Extraction and Crown Prediction Method of Winter Kiwifruit Trees [J]. Smart Agriculture, 2023, 5(4): 92-104. |
Viewed | ||||||
Full text |
|
|||||
Abstract |
|
|||||