基于无人机遥感与改进VMamba-UNet的作物病虫害检测方法

Crop Pests and Diseases Detection Based on UAV Remote Sensing and Improved VMamba-UNet

摘要

针对无人机遥感图像中作物病虫害检测存在的小目标特征弱、背景复杂、传统模型计算效率低且全局建模能力不足的挑战，提出了一种改进的多尺度视觉状态空间U-Net模型（MVMamba-UNet）。该方法通过引入多尺度状态空间模块（MSSM）以线性计算复杂度建模病虫害的长程空间依赖，并集成通道-空间注意力机制（CSA）以自适应聚焦于显著区域、抑制背景干扰，同时结合优化的U型编解码架构实现多尺度特征的高效融合。在自建无人机遥感病虫害数据集上的实验表明，所提模型的精确率、召回率和F1分数分别达到0.942、0.937和0.939，综合性能与训练效率均优于对比模型。研究为田间复杂环境下的作物病虫害精准、实时智能检测提供了有效的解决方案，对推动智慧农业发展具有实际应用价值。

Abstract

In response to the challenges of weak small target features, complex backgrounds, low computational efficiency of traditional models, and insufficient global modeling ability in crop pest and disease detection in unmanned aerial vehicle remote sensing images, an improved multi-scale visual state space U-Net model (MVMamba UNet) is proposed. This method introduces a multi-scale state space module (MSSM) to model the long-range spatial dependencies of pests and diseases with linear computational complexity, and integrates a channel space attention mechanism (CSA) to adaptively focus on salient regions and suppress background interference. At the same time, it combines an optimized U-shaped encoding and decoding architecture to achieve efficient fusion of multi-scale features. The experiments on the self built unmanned aerial vehicle remote sensing pest and disease dataset show that the accuracy, recall, and F1 score of the proposed model reach 0.942, 0.937, and 0.939, respectively, and the comprehensive performance and training efficiency are better than the comparison model. The research provides an effective solution for precise and real-time intelligent detection of crop diseases and pests in complex field environments, which has practical application value for promoting the development of smart agriculture.

关键词

作物病虫害检测 / 无人机遥感 / VMamba-UNet / 改进VMamba-UNet

Key words

Crop pest and disease detection / UAV remote sensing / VMamba-UNet / Improve VMamba UNet

参考文献

[1] 方晨晨, 石繁槐. 基于改进深度残差网络的番茄病害图像识别[J]. 计算机应用, 2020, 40(S1):203-208.

[2] 曾伟辉, 李淼, 李增, 等. 基于高阶残差和参数共享反馈卷积神经网络的农作物病害识别[J]. 电子学报, 2019, 47(09):1979-1986.

[3] 王美华, 吴振鑫, 周祖光. 基于注意力改进CBAM的农作物病虫害细粒度识别研究[J]. 农业机械学报, 2021, 52(04):239-247.

[4] 张善文, 邵彧, 齐国红, 等. 基于多尺度注意力卷积网络的作物害虫检测[J]. 江苏农业学报, 2021, 37(03):579-588.

[5] He H M, Liu L N, Munir S, et al. Crop diversity and pest management in sustainable agriculture. Journal of Integrative Agriculture, 2018, 18(9): 1945-1952. DOI:10.1016/S2095-3119(19)62689-4.

[6] Xavier T W F, Souto R N V, Statella T, et al. Identification of Ramularia leaf blight cotton disease infection levels by multispectral, multiscale UAV Imagery. UAVs, 2019, 3(2): 33-33. DOI:10.3390/UAVs3020033.

[7] Zhang P, Huang Y, Pu R. Monitoring plant diseases and pests through remote sensing technology: A review. Computers and Electronics in Agriculture, 2019, 165, 104943. DOI:10.1016/j.compag.2019.104943

[8] Dehkordi R, Jarroudi M, Kouadio L, et al. Monitoring wheat leaf rust and stripe rust in winter wheat using high resolution UAV based red-green-blue imagery. Remote Sensing, 2020, 12(22): 3696. DOI:10.3390/rs12223696.

[9] Narmilan A, Gonzalez F, Salgadoe A, et al. Detection of White Leaf Disease in Sugarcane Using Machine Learning Techniques over UAV Multispectral Images. UAVs, 2022, 6,230. DOI:10.3390/UAVs6090230

[10] Gao D, Sun Q, Hu B, et al. A framework for agricultural pest and disease monitoring based on internet of things and unmanned aerial vehicles. Sensors, 2020, 20(5): 1487-1487. DOI:10.3390/s20051487.

[11] Amarasingam N, Gonzalez F, Salgadoe A, et al. Detection of White Leaf Disease in Sugarcane Crops Using UAV-Derived RGB Imagery with Existing Deep Learning Models. Remote Sens. 2022, 14,6137.

基金资助

河南省科技攻关项目(252102110343)

河南省校企协同创新项目(26AXQXT120)