Accurate identification of plant diseases is a critical task in agricultural production. The existing deep learning crop disease recognition methods require a large number of labeled images for training, limiting the implementation of large-scale detection. To overcome this limitation, this study explores the application of self-supervised learning (SSL) in plant disease recognition. We propose a new model that combines a masked autoencoder (MAE) and a convolutional block attention module (CBAM) to alleviate the harsh requirements of large amounts of labeled data. The performance of the model was validated on the CCMT dataset and our collected dataset. The results show that the improved model achieves an accuracy of 95.35% and 99.61%, recall of 96.2% and 98.51%, and F1 values of 95.52% and 98.62% on the CCMT dataset and our collected dataset, respectively. Compared with ResNet50, ViT, and MAE, the accuracies on the CCMT dataset improved by 1.2%, 0.7%, and 0.8%, respectively, and the accuracy of our collected dataset improved by 1.3%, 1.6%, and 0.6%, respectively. Through experiments on 21 leaf diseases (early blight, late blight, leaf blight, leaf spot, etc.) of five crops, namely, potato, maize, tomato, cashew, and cassava, our model achieved accurate and rapid detection of plant disease categories. This study provides a reference for research work and engineering applications in crop disease detection.

Soil moisture critically influences crop growth, especially in dryland environments. Precise agricultural management requires real-time monitoring of stratified soil moisture and assessment of crops’ daily water needs. We aim to provide low-cost, high-throughput information acquisition services for dryland regions with underdeveloped infrastructure and offer scientific support for sustainable water resource management. We used UAVs (Unmanned Aerial Vehicles) with multi-spectral sensors for routine maize monitoring, capturing leaf reflectance. Constructing vegetation indices, we quantified the relationship between leaf water content and surface soil moisture, using the Biswas model to predict deep soil moisture distribution. We used UVAs to monitor crop height and calculated the daily water demand for the entire growth period of corn using the Penman Montes equation. We found an R² of 0.8603, RMSE of 2.455%, and MAE of 2.099% between NDVI and canopy leaf water content. A strong linear correlation (R² = 0.7510) between canopy leaf water content and soil moisture was observed in the top 20 cm of soil. Deep soil moisture inversion from the top 20 cm soil moisture showed an R² of 0.9984, with RE mostly under 10%, but exceeding 20% at 120 cm depth. We also constructed a maize height model aligning with a sigmoidal growth curve (R² = 0.9724). Maize’s daily water demand varied from 0.7121 to 9.4263 mm, exhibiting a downward-opening parabolic trend. Integration of rainfall and soil water data allowed for dynamic irrigation adjustments, mitigating drought and water stress effects on crops. We highlighted UAV multi-spectral imaging’s effectiveness in monitoring crop water needs, facilitating quick daily water requirement estimations. Our work offers a scientific foundation for managing maize cultivation in drylands, enhancing water resource utilization.

Soil moisture content is a key factor influencing plant growth and agricultural productivity, directly impacting water uptake, nutrient absorption, and stress resistance. This study proposes a rapid, low-cost, non-destructive method for dynamically monitoring soil moisture at depths of 0–200 cm throughout the crop growth period under dryland conditions, with validation in soybean cultivation. During critical soybean growth stages, UAV multispectral data of the canopy were collected, and ground measurements were conducted for three GPS-referenced 50 cm × 50 cm plots to obtain canopy leaf water content, coverage, and soil volumetric moisture at 20 cm intervals. Ten vegetation indices were constructed from multispectral data to explore statistical relationships between vegetation indices, surface soil moisture, canopy leaf water content, and deeper soil moisture. Predictive models were developed and evaluated. Results showed that the NDVI-based nonlinear regression model achieved the best performance for leaf water content (R² = 0.725), and a significant correlation was found between canopy leaf water content and 0–20 cm soil moisture (R² = 0.705), enabling predictions of deeper soil moisture. Surface soil models accurately estimated 0–200 cm soil moisture distribution (R² = 0.9995). Daily water dynamics simulations provided robust support for precision irrigation management. This study demonstrates that UAV multispectral remote sensing combined with ground sampling is a valuable tool for soybean water management, supporting precision agriculture and sustainable water resource utilization.

Flight altitude is a critical parameter influencing both the spatial resolution and operational efficiency of UAV multispectral imaging; however, its quantitative effects on crop monitoring accuracy remain insufficiently characterized. This study investigated maize in the Yuci District, Jinzhong, China, using multispectral imagery and ground measurements of soil moisture, SPAD, leaf water content (LWC), leaf area index (LAI), plant height (PH), and aboveground biomass (AGB) collected at eight altitudes (65–200 m). Correlation analysis and three modeling approaches were applied: stepwise linear regression (SLR), random forest (RF), and back-propagation neural network (BPNN). Accuracy decreased with altitude. At 65–100 m, the correlations were strongest: LAI–NDVI/GNDVI ranged from 0.818 to 0.938, and SPAD–NDVI/GNDVI exceeded 0.816. At 80–100 m, RMSE values for LAI, SPAD, and LWC were 0.05, 10.37, and 0.67, with RE below 15%. At 200 m, the lowest R² dropped to 0.23, with errors rising sharply. RF and BPNN outperformed SLR, with BPNN yielding the highest accuracy for LAI and AGB. Overall, 65–100 m is optimal for precision monitoring, 120–160 m balances accuracy and efficiency, and 180–200 m suits large-scale reconnaissance. These findings provide methodological guidance for UAV flight parameter optimization in precision agriculture.

Accurate, high-throughput canopy phenotyping using UAV-based multispectral remote sensing is critically important for optimizing the management and breeding of foxtail millet in rainfed regions. This study integrated multi-temporal field measurements of leaf water content, SPAD-derived chlorophyll, and leaf area index (LAI) with UAV imagery (red, green, red-edge, and near-infrared bands) across two sites and two consecutive years (2023 and 2024) in Shanxi Province, China. Various modeling approaches, including Random Forest, Gradient Boosting, and regularized regressions (e.g., Ridge and Lasso), were evaluated for cross-regional and cross-year extrapolation. The results showed that single-site modeling achieved coefficients of determination (R²) of up to 0.95, with mean relative errors of 10–15% in independent validations. When models were transferred between sites, R² generally remained between 0.50 and 0.70, although SPAD estimates exhibited larger deviations under high-nitrogen conditions. Even under severe drought in 2024, cross-year predictions still attained R² values near 0.60. Among these methods, tree-based models demonstrated a strong capability for capturing nonlinear canopy trait dynamics, whereas regularized regressions offered simplicity and interpretability. Incorporating multi-site and multi-year data further enhanced model robustness, increasing R² above 0.80 and markedly reducing average prediction errors. These findings demonstrate that rigorous radiometric calibration and appropriate vegetation index selection enable reliable UAV-based phenotyping for foxtail millet in diverse environments and time frames. Thus, the proposed approach provides strong technical support for precision management and cultivar selection in semi-arid foxtail millet production systems.

Strip intercropping of maize and soybean is a key practice for improving land productivity and ensuring food and oil security in the hilly regions of the Loess Plateau. However, complex interspecific interactions generate highly heterogeneous canopy structures, making it difficult for traditional linear models to capture yield variability within mixed pixels. Based on a single-season (2025) field experiment, this study developed a UAV multispectral imagery-based yield estimation framework integrating multiple machine-learning algorithms. Shapley additive explanations (SHAP) and partial dependence plots (PDP) were used to interpret the spectral–yield relationships under different spatial configurations. The predictive performance of linear regression and eight nonlinear algorithms was compared using 20 spectral features. Ensemble learning outperformed linear approaches in all intercropping scenarios. In the maize–soybean 3:2 pattern, the GBDT model delivered the highest accuracy (R² = 0.849; NRMSE = 9.28%), whereas in the 4:2 pattern with stronger shading stress on soybean, the random forest model showed the greatest robustness (R² = 0.724). Interpretation results indicated that yield in monoculture systems was mainly driven by physiological traits characterized by visible-band indices, while yield in intercropping systems was dominated by structural and stress-response traits represented by near-infrared and soil-adjusted vegetation indices. The generated centimeter-scale yield maps revealed clear strip-like spatial variability driven by interspecific competition. Overall, explainable machine learning combined with UAV multispectral data shows promise for within-season yield estimation in intercropping systems and can support spatially differentiated precision management under the sampled conditions.