《计算机应用》唯一官方网站 ›› 2026, Vol. 46 ›› Issue (10): 3342-3353.DOI: 10.11772/j.issn.1001-9081.2025091184
• 多媒体计算与计算机仿真 • 上一篇
黄会敏1,2, 汪鹏飞1, 赵应1, 袁爱红1,2, 胡瑾1,2, 王建锋3,4(
)
收稿日期:2025-10-11
修回日期:2025-12-11
接受日期:2025-12-12
发布日期:2026-01-04
出版日期:2026-10-10
通讯作者:
王建锋
作者简介:黄会敏(1991—),女,湖北孝感人,讲师,博士,主要研究方向:图像处理、点云配准、小麦条锈病检测基金资助:
Huimin HUANG1,2, Pengfei WANG1, Ying ZHAO1, Aihong YUAN1,2, Jin HU1,2, Jianfeng WANG3,4(
)
Received:2025-10-11
Revised:2025-12-11
Accepted:2025-12-12
Online:2026-01-04
Published:2026-10-10
Contact:
Jianfeng WANG
About author:HUANG Huimin, born in 1991, Ph. D., lecturer. Her research interests include image processing, point cloud registration, wheat stripe rust detection.Supported by:摘要:
针对小麦条锈病多等级识别中背景复杂、病斑尺度差异显著以及相邻类别特征差异细微等问题,提出一种多尺度特征融合与注意力增强残差网络(MSA-ResNet)。首先,利用残差连接缓解深层网络训练中的梯度消失问题,提高模型的特征表达能力;其次,采用自顶向下的多尺度特征融合策略,通过语义对齐增强不同尺度特征的一致性,提高模型对病斑尺寸变化和形态复杂性的适应能力;最后,引入融合通道注意力与空间注意力的卷积块注意力模块(CBAM),增强病斑关键区域的特征响应,并抑制杂草、土壤等背景信息的干扰。在公开数据集Yellow-Rust-19和自建数据集上的实验结果表明,MSA-ResNet的准确率分别达到95.67%和97.17%,相较于EfficientNet-CBAM分别提高了1.67和0.84个百分点。结果表明,MSA-ResNet能够更好地满足实际应用需求,为田间小麦条锈病的多级识别提供更优的解决方案。
中图分类号:
黄会敏, 汪鹏飞, 赵应, 袁爱红, 胡瑾, 王建锋. 基于多尺度特征融合与注意力增强的小麦条锈病识别方法[J]. 计算机应用, 2026, 46(10): 3342-3353.
Huimin HUANG, Pengfei WANG, Ying ZHAO, Aihong YUAN, Jin HU, Jianfeng WANG. Wheat stripe rust recognition method based on multi-scale feature fusion and attention enhancement[J]. Journal of Computer Applications, 2026, 46(10): 3342-3353.
| 层名 | 输入尺寸 | 模块类型 | 输出 尺寸 | 通道 数 | 重复 次数 |
|---|---|---|---|---|---|
| Input | 224×224×3 | Input layer | 224×224 | 3 | 1 |
| Stem | 224×224×3 | Conv(7×7) | 56×56 | 64 | 1 |
| Layer1 | 56×56×64 | ResNet block×3 | 56×56 | 256 | 3 |
| Layer2 | 56×56×256 | ResNet block×4 | 28×28 | 512 | 4 |
| Layer3 | 28×28×512 | ResNet block×6 | 14×14 | 1 024 | 6 |
| Layer4 | 14×14×1 024 | ResNet block×3 | 7×7 | 2 048 | 3 |
| Lateral conv1 | 28×28×512 | UFPM block | 28×28 | 256 | 1 |
| Lateral conv2 | 14×14×1 024 | UFPM block | 28×28 | 256 | 1 |
| Lateral conv3 | 7×7×2 048 | UFPM block | 28×28 | 256 | 1 |
| Fusion convs | 28×28×256 | UFPM block | 28×28 | 256 | 3 |
Channel Attention | 28×28×256 | CBAM block | 28×28 | 256 | 1 |
Spatial Attention | 28×28×256 | CBAM block | 28×28 | 256 | 1 |
| Global Pooling | 28×28×256 | Adaptive AvgPool(1×1) | 1×1 | 256 | 1 |
| FC | 1×1×256 | Linear(256→6) | 1×1×6 | 6 | 1 |
表1 MSA-ResNet的主要参数
Tab. 1 Main parameters of MSA-ResNet
| 层名 | 输入尺寸 | 模块类型 | 输出 尺寸 | 通道 数 | 重复 次数 |
|---|---|---|---|---|---|
| Input | 224×224×3 | Input layer | 224×224 | 3 | 1 |
| Stem | 224×224×3 | Conv(7×7) | 56×56 | 64 | 1 |
| Layer1 | 56×56×64 | ResNet block×3 | 56×56 | 256 | 3 |
| Layer2 | 56×56×256 | ResNet block×4 | 28×28 | 512 | 4 |
| Layer3 | 28×28×512 | ResNet block×6 | 14×14 | 1 024 | 6 |
| Layer4 | 14×14×1 024 | ResNet block×3 | 7×7 | 2 048 | 3 |
| Lateral conv1 | 28×28×512 | UFPM block | 28×28 | 256 | 1 |
| Lateral conv2 | 14×14×1 024 | UFPM block | 28×28 | 256 | 1 |
| Lateral conv3 | 7×7×2 048 | UFPM block | 28×28 | 256 | 1 |
| Fusion convs | 28×28×256 | UFPM block | 28×28 | 256 | 3 |
Channel Attention | 28×28×256 | CBAM block | 28×28 | 256 | 1 |
Spatial Attention | 28×28×256 | CBAM block | 28×28 | 256 | 1 |
| Global Pooling | 28×28×256 | Adaptive AvgPool(1×1) | 1×1 | 256 | 1 |
| FC | 1×1×256 | Linear(256→6) | 1×1×6 | 6 | 1 |
| 等级 | Yellow-Rust-19 | 自建数据集 | ||
|---|---|---|---|---|
| 采集样本数 | 增强后样本数 | 采集样本数 | 增强后样本数 | |
| 合计 | 5 421 | 15 000 | 1 787 | 6 000 |
| IM | 205 | 2 500 | 180 | 1 000 |
| R | 361 | 2 500 | 407 | 1 000 |
| MR | 564 | 2 500 | 443 | 1 000 |
| MRMS | 1 135 | 2 500 | 409 | 1 000 |
| MS | 1 795 | 2 500 | 173 | 1 000 |
| S | 1 361 | 2 500 | 175 | 1 000 |
表2 两个数据集的样本分布
Tab. 2 Sample distribution of two datasets
| 等级 | Yellow-Rust-19 | 自建数据集 | ||
|---|---|---|---|---|
| 采集样本数 | 增强后样本数 | 采集样本数 | 增强后样本数 | |
| 合计 | 5 421 | 15 000 | 1 787 | 6 000 |
| IM | 205 | 2 500 | 180 | 1 000 |
| R | 361 | 2 500 | 407 | 1 000 |
| MR | 564 | 2 500 | 443 | 1 000 |
| MRMS | 1 135 | 2 500 | 409 | 1 000 |
| MS | 1 795 | 2 500 | 173 | 1 000 |
| S | 1 361 | 2 500 | 175 | 1 000 |
| 模型 | 独立测试集 | 5-折交叉验证 | ||||||
|---|---|---|---|---|---|---|---|---|
| 精准率 | 召回率 | F1值 | 准确率 | 精准率(均值±σ) | 召回率(均值±σ) | F1值(均值±σ) | 准确率(均值±σ) | |
| MobileNetV2[ | 92.52 | 92.40 | 92.41 | 92.40 | 92.15±0.45 | 92.08±0.43 | 92.11±0.44 | 92.08±0.43 |
| SE-Net[ | 93.96 | 94.00 | 93.96 | 94.00 | 93.72±0.39 | 93.65±0.37 | 93.68±0.38 | 93.65±0.37 |
| ViT[ | 93.66 | 93.67 | 93.65 | 93.67 | 93.35±0.42 | 93.28±0.40 | 93.31±0.41 | 93.28±0.40 |
| CNN-CGLCM_HSV[ | 92.43 | 92.40 | 92.36 | 92.40 | — | — | — | — |
| S-ResNet34[ | 93.25 | 93.27 | 93.24 | 93.27 | 93.01±0.52 | 92.95±0.48 | 92.98±0.50 | 92.95±0.48 |
| Imp-DenseNet[ | 93.85 | 93.80 | 93.73 | 93.80 | 93.56±0.43 | 93.50±0.40 | 93.52±0.41 | 93.50±0.40 |
| EfficientNet-CBAM[ | 94.02 | 94.00 | 93.97 | 94.00 | 93.76±0.45 | 93.68±0.41 | 93.72±0.43 | 93.68±0.41 |
| Mobile-DMSANet[ | 93.39 | 93.47 | 93.40 | 93.47 | 93.12±0.49 | 93.08±0.46 | 93.10±0.47 | 93.08±0.46 |
| ConMamba[ | 92.96 | 92.80 | 92.81 | 92.80 | 92.85±0.42 | 92.70±0.40 | 92.75±0.41 | 92.70±0.40 |
| HyperKAN[ | 93.41 | 93.47 | 93.43 | 93.47 | 93.30±0.45 | 93.35±0.43 | 93.32±0.44 | 93.35±0.43 |
| MSA-ResNet | 95.65 | 95.67 | 95.66 | 95.67 | 95.42±0.31 | 95.38±0.27 | 95.40±0.29 | 95.38±0.27 |
表3 Yellow-Rust-19数据集上各方法的识别性能对比(5-折交叉验证) (%)
Tab. 3 Comparison of recognition performance of various methods on Yellow-Rust-19 dataset (5-fold cross-validation)
| 模型 | 独立测试集 | 5-折交叉验证 | ||||||
|---|---|---|---|---|---|---|---|---|
| 精准率 | 召回率 | F1值 | 准确率 | 精准率(均值±σ) | 召回率(均值±σ) | F1值(均值±σ) | 准确率(均值±σ) | |
| MobileNetV2[ | 92.52 | 92.40 | 92.41 | 92.40 | 92.15±0.45 | 92.08±0.43 | 92.11±0.44 | 92.08±0.43 |
| SE-Net[ | 93.96 | 94.00 | 93.96 | 94.00 | 93.72±0.39 | 93.65±0.37 | 93.68±0.38 | 93.65±0.37 |
| ViT[ | 93.66 | 93.67 | 93.65 | 93.67 | 93.35±0.42 | 93.28±0.40 | 93.31±0.41 | 93.28±0.40 |
| CNN-CGLCM_HSV[ | 92.43 | 92.40 | 92.36 | 92.40 | — | — | — | — |
| S-ResNet34[ | 93.25 | 93.27 | 93.24 | 93.27 | 93.01±0.52 | 92.95±0.48 | 92.98±0.50 | 92.95±0.48 |
| Imp-DenseNet[ | 93.85 | 93.80 | 93.73 | 93.80 | 93.56±0.43 | 93.50±0.40 | 93.52±0.41 | 93.50±0.40 |
| EfficientNet-CBAM[ | 94.02 | 94.00 | 93.97 | 94.00 | 93.76±0.45 | 93.68±0.41 | 93.72±0.43 | 93.68±0.41 |
| Mobile-DMSANet[ | 93.39 | 93.47 | 93.40 | 93.47 | 93.12±0.49 | 93.08±0.46 | 93.10±0.47 | 93.08±0.46 |
| ConMamba[ | 92.96 | 92.80 | 92.81 | 92.80 | 92.85±0.42 | 92.70±0.40 | 92.75±0.41 | 92.70±0.40 |
| HyperKAN[ | 93.41 | 93.47 | 93.43 | 93.47 | 93.30±0.45 | 93.35±0.43 | 93.32±0.44 | 93.35±0.43 |
| MSA-ResNet | 95.65 | 95.67 | 95.66 | 95.67 | 95.42±0.31 | 95.38±0.27 | 95.40±0.29 | 95.38±0.27 |
| 模型 | 独立测试集 | 5-折交叉验证 | ||||||
|---|---|---|---|---|---|---|---|---|
| 精准率 | 召回率 | F1值 | 准确率 | 精准率(均值±σ) | 召回率(均值±σ) | F1值(均值±σ) | 准确率(均值±σ) | |
| MobileNetV2[ | 94.28 | 94.17 | 94.18 | 94.17 | 93.92±0.41 | 93.85±0.39 | 93.88±0.40 | 93.85±0.39 |
| SE-Net[ | 95.53 | 95.50 | 95.49 | 95.50 | 95.23±0.36 | 95.18±0.33 | 95.20±0.34 | 95.18±0.33 |
| ViT[ | 94.98 | 95.00 | 94.96 | 95.00 | 94.72±0.38 | 94.65±0.35 | 94.68±0.36 | 94.65±0.35 |
| CNN-CGLCM_HSV[ | — | — | — | — | — | — | — | — |
| S-ResNet34[ | 95.97 | 95.83 | 95.83 | 95.83 | 95.62±0.38 | 95.57±0.35 | 95.59±0.36 | 95.57±0.35 |
| Imp-DenseNet[ | 96.27 | 96.17 | 96.16 | 96.17 | 95.89±0.34 | 95.85±0.31 | 95.86±0.32 | 95.85±0.31 |
| EfficientNet-CBAM[ | 96.37 | 96.33 | 96.32 | 96.33 | 96.05±0.33 | 96.01±0.28 | 96.02±0.30 | 96.01±0.28 |
| Mobile-DMSANet[ | 96.19 | 96.17 | 96.15 | 96.17 | 95.91±0.32 | 95.87±0.29 | 95.88±0.30 | 95.87±0.29 |
| ConMamba[ | 96.73 | 96.67 | 96.67 | 96.67 | 96.50±0.30 | 96.40±0.28 | 96.45±0.29 | 96.40±0.28 |
| HyperKAN[ | 95.82 | 95.83 | 95.81 | 95.83 | 95.50±0.35 | 95.60±0.33 | 95.55±0.34 | 95.60±0.33 |
| MSA-ResNet | 97.19 | 97.17 | 97.17 | 97.17 | 97.03±0.22 | 96.98±0.19 | 96.99±0.20 | 96.98±0.19 |
表4 自建数据集上各模型的识别性能对比(5-折交叉验证) (%)
Tab. 4 Comparison of recognition performance of various models on self-built dataset (5-fold cross-validation)
| 模型 | 独立测试集 | 5-折交叉验证 | ||||||
|---|---|---|---|---|---|---|---|---|
| 精准率 | 召回率 | F1值 | 准确率 | 精准率(均值±σ) | 召回率(均值±σ) | F1值(均值±σ) | 准确率(均值±σ) | |
| MobileNetV2[ | 94.28 | 94.17 | 94.18 | 94.17 | 93.92±0.41 | 93.85±0.39 | 93.88±0.40 | 93.85±0.39 |
| SE-Net[ | 95.53 | 95.50 | 95.49 | 95.50 | 95.23±0.36 | 95.18±0.33 | 95.20±0.34 | 95.18±0.33 |
| ViT[ | 94.98 | 95.00 | 94.96 | 95.00 | 94.72±0.38 | 94.65±0.35 | 94.68±0.36 | 94.65±0.35 |
| CNN-CGLCM_HSV[ | — | — | — | — | — | — | — | — |
| S-ResNet34[ | 95.97 | 95.83 | 95.83 | 95.83 | 95.62±0.38 | 95.57±0.35 | 95.59±0.36 | 95.57±0.35 |
| Imp-DenseNet[ | 96.27 | 96.17 | 96.16 | 96.17 | 95.89±0.34 | 95.85±0.31 | 95.86±0.32 | 95.85±0.31 |
| EfficientNet-CBAM[ | 96.37 | 96.33 | 96.32 | 96.33 | 96.05±0.33 | 96.01±0.28 | 96.02±0.30 | 96.01±0.28 |
| Mobile-DMSANet[ | 96.19 | 96.17 | 96.15 | 96.17 | 95.91±0.32 | 95.87±0.29 | 95.88±0.30 | 95.87±0.29 |
| ConMamba[ | 96.73 | 96.67 | 96.67 | 96.67 | 96.50±0.30 | 96.40±0.28 | 96.45±0.29 | 96.40±0.28 |
| HyperKAN[ | 95.82 | 95.83 | 95.81 | 95.83 | 95.50±0.35 | 95.60±0.33 | 95.55±0.34 | 95.60±0.33 |
| MSA-ResNet | 97.19 | 97.17 | 97.17 | 97.17 | 97.03±0.22 | 96.98±0.19 | 96.99±0.20 | 96.98±0.19 |
UFPM 输入特征层 | 注意力 | 独立测试集 | 5-折交叉验证 | ||||||
|---|---|---|---|---|---|---|---|---|---|
| 精准率 | 召回率 | F1值 | 准确率 | 精准率(均值±σ) | 召回率(均值±σ) | F1值(均值±σ) | 准确率(均值±σ) | ||
| × | × | 94.07 | 94.07 | 94.06 | 94.07 | 93.82±0.35 | 93.78±0.32 | 93.80±0.33 | 93.78±0.32 |
| C2, C3, C4 | × | 94.76 | 94.80 | 94.78 | 94.80 | 94.52±0.29 | 94.48±0.27 | 94.50±0.28 | 94.48±0.27 |
| × | √ | 94.26 | 94.27 | 94.26 | 94.26 | 94.01±0.33 | 93.97±0.30 | 93.99±0.31 | 93.97±0.30 |
| C2, C3 | √ | 94.30 | 94.20 | 94.17 | 94.20 | 94.05±0.34 | 93.95±0.31 | 93.99±0.32 | 93.95±0.31 |
| C2, C3, C4 | √ | 95.65 | 95.67 | 95.66 | 95.67 | 95.42±0.31 | 95.38±0.27 | 95.40±0.29 | 95.38±0.27 |
表5 Yellow-Rust-19数据集上多尺度特征融合与注意力机制的消融实验结果(5-折交叉验证) (%)
Tab. 5 Ablation study results of multi-scale feature fusion and attention mechanisms on Yellow-Rust-19 dataset (5-fold cross-validation)
UFPM 输入特征层 | 注意力 | 独立测试集 | 5-折交叉验证 | ||||||
|---|---|---|---|---|---|---|---|---|---|
| 精准率 | 召回率 | F1值 | 准确率 | 精准率(均值±σ) | 召回率(均值±σ) | F1值(均值±σ) | 准确率(均值±σ) | ||
| × | × | 94.07 | 94.07 | 94.06 | 94.07 | 93.82±0.35 | 93.78±0.32 | 93.80±0.33 | 93.78±0.32 |
| C2, C3, C4 | × | 94.76 | 94.80 | 94.78 | 94.80 | 94.52±0.29 | 94.48±0.27 | 94.50±0.28 | 94.48±0.27 |
| × | √ | 94.26 | 94.27 | 94.26 | 94.26 | 94.01±0.33 | 93.97±0.30 | 93.99±0.31 | 93.97±0.30 |
| C2, C3 | √ | 94.30 | 94.20 | 94.17 | 94.20 | 94.05±0.34 | 93.95±0.31 | 93.99±0.32 | 93.95±0.31 |
| C2, C3, C4 | √ | 95.65 | 95.67 | 95.66 | 95.67 | 95.42±0.31 | 95.38±0.27 | 95.40±0.29 | 95.38±0.27 |
UFPM 输入特征层 | 注意力 | 独立测试集 | 5-折交叉验证 | ||||||
|---|---|---|---|---|---|---|---|---|---|
| 精准率 | 召回率 | F1值 | 准确率 | 精准率(均值±σ) | 召回率(均值±σ) | F1值(均值±σ) | 准确率(均值±σ) | ||
| × | × | 95.13 | 95.00 | 95.01 | 95.00 | 94.85±0.28 | 94.80±0.25 | 94.82±0.26 | 94.80±0.25 |
| C2, C3, C4 | × | 96.86 | 96.83 | 96.81 | 96.83 | 96.62±0.21 | 96.58±0.19 | 96.60±0.20 | 96.58±0.19 |
| × | √ | 96.34 | 96.33 | 96.32 | 96.33 | 96.08±0.24 | 96.03±0.22 | 96.05±0.23 | 96.03±0.22 |
| C2, C3 | √ | 96.03 | 96.00 | 95.98 | 96.00 | 95.76±0.26 | 95.70±0.24 | 95.73±0.25 | 95.70±0.24 |
| C2, C3, C4 | √ | 97.19 | 97.17 | 97.17 | 97.17 | 97.03±0.22 | 96.98±0.19 | 96.99±0.20 | 96.98±0.19 |
表6 自建数据集上多尺度特征融合与注意力机制的消融实验结果(5-折交叉验证) (%)
Tab. 6 Ablation study results of multi-scale feature fusion and attention mechanisms on self-built dataset (5-fold cross-validation)
UFPM 输入特征层 | 注意力 | 独立测试集 | 5-折交叉验证 | ||||||
|---|---|---|---|---|---|---|---|---|---|
| 精准率 | 召回率 | F1值 | 准确率 | 精准率(均值±σ) | 召回率(均值±σ) | F1值(均值±σ) | 准确率(均值±σ) | ||
| × | × | 95.13 | 95.00 | 95.01 | 95.00 | 94.85±0.28 | 94.80±0.25 | 94.82±0.26 | 94.80±0.25 |
| C2, C3, C4 | × | 96.86 | 96.83 | 96.81 | 96.83 | 96.62±0.21 | 96.58±0.19 | 96.60±0.20 | 96.58±0.19 |
| × | √ | 96.34 | 96.33 | 96.32 | 96.33 | 96.08±0.24 | 96.03±0.22 | 96.05±0.23 | 96.03±0.22 |
| C2, C3 | √ | 96.03 | 96.00 | 95.98 | 96.00 | 95.76±0.26 | 95.70±0.24 | 95.73±0.25 | 95.70±0.24 |
| C2, C3, C4 | √ | 97.19 | 97.17 | 97.17 | 97.17 | 97.03±0.22 | 96.98±0.19 | 96.99±0.20 | 96.98±0.19 |
| [1] | Bukhari H R, Mumtaz R, Inayat S, et al. Assessing the impact of segmentation on wheat stripe rust disease classification using computer vision and deep learning[J]. IEEE Access, 2021, 9: 164986-165004. |
| [2] | University of Nebraska-Lincoln. Wheat disease identification[EB/OL]. [2025-07-10].. |
| [3] | Martinelli F, Scalenghe R, Davino S, et al. Advanced methods of plant disease detection. a review[J]. Agronomy for Sustainable Development, 2015, 35(1): 1-25. |
| [4] | Shinde S S, Kulkarni M. Review paper on prediction of crop disease using IoT and machine learning[C]// ICTEE 2017. Piscataway: IEEE, 2017: 1-4. |
| [5] | Deng J, Hong D, Li C, et al. RustQNet: multimodal deep learning for quantitative inversion of wheat stripe rust disease index[J]. Computers and Electronics in Agriculture, 2024, 225: No.109245. |
| [6] | 朱文静,陈华,李林,等. 基于红外热成像边缘检测算法的小麦叶锈病分级研究[J]. 农业机械学报, 2019, 50(4): 36-41, 48. |
| Zhu Wenjing, Chen Hua, Li Lin, et al. Grading of wheat leaf rust based on edge detection of infrared thermal imaging[J]. Transactions of the Chinese Society for Agricultural Machinery, 2019, 50(4): 36-41, 48. | |
| [7] | 蒋小敏,冯海宽,常红,等. 基于数码图像的小麦条锈病病害程度分级方法[J]. 江苏农业科学, 2021, 49(23): 109-115. |
| Jiang Xiaomin, Feng Haikuan, Chang Hong, et al. Grade classification method of wheat stripe rust diseases based on digital images[J]. Jiangsu Agricultural Sciences, 2021, 49(23): 109-115. | |
| [8] | Kumar S, Singh R, Kumar S, et al. Enhancing wheat yellow rust detection through modified deep learning approach[J]. Current Plant Biology, 2025, 42: No.100472. |
| [9] | Tola F G, Blitti K E K, Diwan A, et al. Automated wheat rust disease classification using convolutional neural networks with transfer learning[C]// ENSYMP 2024. Piscataway: IEEE, 2024: 1-6. |
| [10] | Tang Z, Wang M, Schirrmann M, et al. Affordable high throughput field detection of wheat stripe rust using deep learning with semi-automated image labeling[J]. Computers and Electronics in Agriculture, 2023, 207: No.107709. |
| [11] | 毛锐,张宇晨,王泽玺,等. 利用改进Faster-RCNN识别小麦条锈病和黄矮病[J]. 农业工程学报, 2022, 38(17): 176-185. |
| Mao Rui, Zhang Yuchen, Wang Zexi, et al. Recognizing stripe rust and yellow dwarf of wheat using improved Faster-RCNN[J]. Transactions of the Chinese Society of Agricultural Engineering, 2022, 38(17): 176-185. | |
| [12] | Saleem S, Hussain A, Majeed N, et al. A multi-scale feature extraction and fusion deep learning method for classification of wheat diseases[J]. Journal of Computer Science, 2025, 21(1): 34-42. |
| [13] | Hassan A, Mumtaz R, Mahmood Z, et al. Wheat leaf localization and segmentation for yellow rust disease detection in complex natural backgrounds[J]. Alexandria Engineering Journal, 2024, 107: 786-798. |
| [14] | Pan Q, Gao M, Wu P, et al. Image classification of wheat rust based on ensemble learning[J]. Sensors, 2022, 22(16): No.6047. |
| [15] | Hayit T, Erbay H, Varçın F, et al. Determination of the severity level of yellow rust disease in wheat by using convolutional neural networks[J]. Journal of Plant Pathology, 2021, 103(3): 923-934. |
| [16] | Hayıt T, Erbay H, Varçın F, et al. The classification of wheat yellow rust disease based on a combination of textural and deep features[J]. Multimedia Tools and Applications, 2023, 82(30): 47405-47423. |
| [17] | 郭伟,党梦佳,贾箫,等. 基于深度学习的小麦条锈病病害等级识别[J]. 华南农业大学学报, 2023, 44(4): 604-612. |
| Guo Wei, Dang Mengjia, Jia Xiao, et al. Grade classification of wheat stripe rust disease based on deep learning[J]. Journal of South China Agricultural University, 2023, 44(4): 604-612. | |
| [18] | Syeda F, Jameel A, Alani N, et al. Automated detection and severity prediction of wheat rust using cost-effective Xception architecture[J]. Plant, Cell and Environment, 2025, 48(6): 4126-4139. |
| [19] | Hu J, Shen L, Sun G. Squeeze-and-excitation networks[C]// CVPR 2018. Piscataway: IEEE, 2018: 7132-7141. |
| [20] | Wang Q, Wu B, Zhu P, et al. ECA-Net: efficient channel attention for deep convolutional neural networks[C]// CVPR 2020. Piscataway: IEEE, 2020: 11531-11539. |
| [21] | 鲍文霞,赵诗意,黄林生,等. 基于轻量级密集多尺度注意力网络的小麦叶部锈病识别方法[J]. 农业机械学报, 2024, 55(11): 21-31. |
| Bao Wenxia, Zhao Shiyi, Huang Linsheng, et al. Lightweight dense multi-scale attention network for identification of rust on wheat leaves[J]. Transactions of the Chinese Society for Agricultural Machinery, 2024, 55(11): 21-31. | |
| [22] | Li S, Liu W, Ji M, et al. GLNet: global-local feature network for wheat leaf disease image classification[J]. Frontiers in Plant Science, 2024, 15: No.1471705. |
| [23] | 尉国帅,贺佳,常宝方,等. 基于改进S-ResNet34模型的小麦条锈病等级识别研究[J]. 南京农业大学学报, 2025, 48(1): 230-239. |
| Wei Guoshuai, He Jia, Chang Baofang, et al. Classification of wheat stripe rust based on improved S-ResNet34 model[J]. Journal of Nanjing Agricultural University, 2025, 48(1): 230-239. | |
| [24] | Szegedy C, Liu W, Jia Y, et al. Going deeper with convolutions[C]// CVPR 2015. Piscataway: IEEE, 2015: 1-9. |
| [25] | Chang S, Yang G, Cheng J, et al. Recognition of wheat rusts in a field environment based on improved DenseNet[J]. Biosystems Engineering, 2024, 238: 10-21. |
| [26] | Lin T Y, Dollár P, Girshick R, et al. Feature pyramid networks for object detection[C]// CVPR 2017. Piscataway: IEEE, 2017: 936-944. |
| [27] | Nigam S, Jain R, Singh V K. Automated severity level estimation of wheat rust using an EfficientNet-CBAM hybrid model[J]. Frontiers in Plant Science, 2025, 16: No.1540642. |
| [28] | Woo S, Park J, Lee J Y, et al. CBAM: convolutional block attention module[C]// ECCV 2018, LNCS 11211. Cham: Springer, 2018: 3-19. |
| [29] | Al Mamun A, Zhang M, Ahmedt-Aristizabal D, et al. ConMamba: contrastive vision Mamba for plant disease detection[J]. Pattern Recognition, 2026, 176: No.113177. |
| [30] | Fang X, Wang B, Huan C, et al. HyperKAN: hypergraph representation learning with Kolmogorov-Arnold networks[C]// ICASSP 2025. Piscataway: IEEE, 2025: 1-5. |
| [31] | He K, Zhang X, Ren S, et al. Deep residual learning for image recognition[C]// CVPR 2016. Piscataway: IEEE, 2016: 770-778. |
| [32] | Sandler M, Howard A, Zhu M, et al. MobileNetV2: inverted residuals and linear bottlenecks[C]// CVPR 2018. Piscataway: IEEE, 2018: 4510-4520. |
| [33] | Dosovitskiy A, Beyer L, Kolesnikov A, et al. An image is worth 16x16 words: Transformers for image recognition at scale[PP/OL]. V2. arXiv (2021-06-03) [2025-07-10].. |
| [1] | 季长清, 张紫滢, 汪祖民. 基于增强图神经网络的多尺度Transformer的连续无创血压预测方法[J]. 《计算机应用》唯一官方网站, 2026, 46(9): 3043-3053. |
| [2] | 欧宇枫, 王嘉宏. 基于双分支特征提取的印刷电路板小目标缺陷检测方法[J]. 《计算机应用》唯一官方网站, 2026, 46(9): 2977-2986. |
| [3] | 陈江彦, 王彦丹, 刘翼虎, 马应龙. 基于协同查询优化的长尾目标检测框架[J]. 《计算机应用》唯一官方网站, 2026, 46(9): 2769-2775. |
| [4] | 谢丽霞, 曹超越, 杨宏宇. 基于改进DeepLabV3+的ADS-B攻击检测方法[J]. 《计算机应用》唯一官方网站, 2026, 46(9): 2867-2876. |
| [5] | 王伊璠, 韩虎, 李栋, 范雅婷, 李琳. 基于异构增强与多源知识融合的方面级情感分析[J]. 《计算机应用》唯一官方网站, 2026, 46(9): 2752-2760. |
| [6] | 刘威, 李维刚, 田志强. 面向点云分类与分割的层次化旋转不变几何结构的表征学习方法[J]. 《计算机应用》唯一官方网站, 2026, 46(8): 2594-2602. |
| [7] | 白杰龙, 方晨韵, 乔志伟. 基于多尺度注意力自适应融合的稀疏CT伪影抑制Transformer网络[J]. 《计算机应用》唯一官方网站, 2026, 46(8): 2660-2667. |
| [8] | 李光辉, 屈立成. 基于改进Lite-Mono架构的动态空洞卷积和分层注意力模型[J]. 《计算机应用》唯一官方网站, 2026, 46(8): 2567-2576. |
| [9] | 黄远航, 荣娜. 结合注意力机制与深度强化学习的无模型光伏接入容量评估方法[J]. 《计算机应用》唯一官方网站, 2026, 46(8): 2708-2715. |
| [10] | 黄麟, 李英顺, 佟维妍, 张树园, 王子涵. 基于改进实时检测Transformer的遥感图像目标检测算法[J]. 《计算机应用》唯一官方网站, 2026, 46(8): 2652-2659. |
| [11] | 郭晓金, 隋旭洋, 周柯男. 基于深度学习的STAR⁃RIS辅助混合场通信系统信道估计[J]. 《计算机应用》唯一官方网站, 2026, 46(8): 2548-2554. |
| [12] | 刘明, 沈东奇, 孟子洋. 双分支结构下多层次特征融合的点云配准网络[J]. 《计算机应用》唯一官方网站, 2026, 46(8): 2584-2593. |
| [13] | 刘凤春, 邵馨莹, 张春英, 王立亚, 任静. 多尺度特征优化的单目深度估计框架FCMdepth[J]. 《计算机应用》唯一官方网站, 2026, 46(8): 2603-2611. |
| [14] | 颜建强, 董贝贝, 曲博婷, 彭晨. 融合多源信息与图级注意力的双向扩散动态图卷积交通流预测网络[J]. 《计算机应用》唯一官方网站, 2026, 46(7): 2327-2333. |
| [15] | 刘雯玲, 李勇, 李佳慧, 张厚康. 面向不均匀雾霾遥感图像的无监督去雾方法[J]. 《计算机应用》唯一官方网站, 2026, 46(7): 2297-2306. |
| 阅读次数 | ||||||
|
全文 |
|
|||||
|
摘要 |
|
|||||