The segmentation of plant disease images enables researchers to quantify the proportion of disease spots on leaves, known as disease severity. Current deep learning methods predominantly focus on single diseases, simple lesions, or laboratory-controlled environments. In this study, we established and publicly released image datasets of field scenarios for three diseases: soybean bacterial blight (SBB), wheat stripe rust (WSR), and cedar apple rust (CAR). We developed Plant Disease Segmentation Networks (PDSNets) based on LinkNet with ResNet-18 as the encoder, including three versions: ×1.0, ×0.75, and ×0.5. The ×1.0 version incorporates a 4 × 4 embedding layer to enhance prediction speed, while versions ×0.75 and ×0.5 are lightweight variants with reduced channel numbers within the same architecture. Their parameter counts are 11.53 M, 6.50 M, and 2.90 M, respectively. PDSNetx0.5 achieved an overall F1 score of 91.96%, an Intersection over Union (IoU) of 85.85% for segmentation, and a coefficient of determination (R²) of 0.908 for severity estimation. On a local central processing unit (CPU), PDSNetx0.5 demonstrated a prediction speed of 34.18 images (640 × 640 pixels) per second, which is 2.66 times faster than LinkNet. Our work provides an efficient and automated approach for assessing plant disease severity in field scenarios.

A rapid and accurate determination of large-scale winter wheat yield is significant for food security and policy formulation. In this study, meteorological data and enhanced vegetation index (EVI) were used to estimate the winter wheat yield in Henan Province, China, by constructing a deep learning model. The deep learning model combines CNN feature extraction and makes full use of the sequence data processing capability of the LSTM and a multi-head attention mechanism to develop a novel CNN–MALSTM estimation model, which can capture the information of input sequences in different feature subspaces to enhance the expressiveness of the model. A CNN–LSTM baseline model was also constructed for comparison. Compared with the baseline model (R² = 0.75, RMSE = 646.53 kg/ha, and MAPE = 8.82%), the proposed CNN–MALSTM model (R² = 0.79, RMSE = 576.01 kg/ha, MAPE = 7.29%) could more accurately estimate the yield. Based on the cross-validation with one year of left-out data and the input of the fertility period by fertility period to explore the sensitivity of the model to data from different fertility periods to the final yield, an annual yield distribution map of Henan Province was constructed. Through cross-validation, the stability of the model in different years was assessed. The results showed that the model could obtain the best prediction of the yield approximately 20 days in advance. In terms of the spatial distribution of the yield in Henan Province on a yearly basis, the estimated yield showed an overall uptrend from west to east, consistent with the trend in the statistical yearbook of the yield for Henan Province. Thus, it can be concluded that the proposed CNN–MALSTM model can provide stable yield estimation results.