Detecting pests and diseases on maize leaves is challenging. This is especially true under complex conditions, such as variable lighting and occlusion. Current methods suffer from low detection accuracy. They also lack sufficient real-time performance. Hence, this study introduces the lightweight detection method YOLOv11-RCDWD based on an improved YOLOv11 model. The proposed approach enhances the YOLOv11 model by incorporating the RepLKNet module as the backbone, which significantly enhances the model’s capacity to capture characteristics of maize leaf pests and diseases. Additionally, the CBAM is embedded within the neck feature extraction network to further refine the feature representation to augment the model’s capability to identify and select essential features by introducing attention mechanisms in both the channel and spatial dimensions, thereby improving the accuracy of feature expression. We have also improved the model by incorporating the DynamicHead module, WIoU loss function, and DynamicATSS label assignment strategy, which collectively enhance detection accuracy, efficiency, and robustness through optimized attention mechanisms, better handling of low-quality samples, and dynamic sample selection during training. The experimental findings indicate that the improved YOLOv11-RCDWD model effectively detected pests and diseases on maize leaves. The precision reached 92.6%, while the recall was 85.4%. The F1 score was 88.9%, and the mAP@0.5 and mAP@0.5~0.95 demonstrated an improvement of 4.9% and 9.0% over the baseline YOLOv11s. Notably, the YOLOv11-RCDWD model significantly outperformed other architectures such as Faster R-CNN, SSD, and various models within the YOLO series, demonstrating superior capabilities in terms of detection speed, parameter count, computational efficiency, and memory utilization. This model achieves an optimal balance between detection performance and resource efficiency. Overall, the improved YOLOv11-RCDWD model significantly reduces detection time and memory usage while maintaining high detection accuracy, supporting the automated detection of maize pests and diseases, and offering a robust solution for intelligent monitoring of agricultural pests.

Detecting pests and diseases on maize leaves under complex conditions, such as varying lighting and occlusion, is challenging, suffering from low detection accuracy and insufficient real-time performance. Hence, this study introduces the lightweight detection method YOLOv11-RepLKNet-CBAM-DynamicHead-WIoU-DynamicATSS (YOLOv11-RCDWD) based on an improved YOLOv11 model. The proposed method builds on the YOLOv11 model by introducing the RepLKNet module as the backbone network, which enhances the model's ability to express features of maize leaf pests and diseases while reducing the network parameters and computational complexity due to its efficient feature extraction capabilities. The Convolutional Block Attention Module (CBAM) is integrated into the neck feature extraction network to enhance the model's ability to select key features by introducing attention mechanisms in both the channel and spatial dimensions, thereby improving feature expression accuracy. Additionally, the DynamicHead detection head in the feature fusion structure dynamically adjusts the focus of attention through a unified attention mechanism for scale, space, and task awareness, improving detection accuracy and efficiency. Furthermore, the Weighted IoU (WIoU) loss function effectively handles the impact of low-quality samples on bounding box regression through a dynamic non-monotonic focusing mechanism. WIoU reduces the competitiveness of high-quality anchor boxes and the harmful gradients generated by low-quality samples, thereby enhancing the model's localization performance. Finally, the Dynamic Adaptive Training Sample Selection (DynamicATSS) label assignment strategy optimizes sample assignment based on statistical information during training by dynamically adjusting the selection mechanism for positive and negative samples, further enhancing the model's detection capabilities. Experimental results demonstrate that the improved YOLOv11-RCDWD model effectively detects pests and diseases on maize leaves. The precision, recall, and F1 score reach 92.6%, 85.4%, and 88.9%, respectively, an improvement of 4.9% and 9.0% over the baseline YOLOv11s. Notably, YOLOv11-RCDWD outperforms models such as Faster R-CNN, SSD, YOLOv5, YOLOv7, YOLOv8n, YOLOv9t, and YOLOv10n in detection speed (0.035 s), number of parameters (0.81 M), computational load (27.23 GFLOPs), and memory usage (6.46 MB), achieving the best balance between detection performance and resource efficiency. Overall, the improved YOLOv11-RCDWD model significantly reduces detection time and memory usage while maintaining high detection accuracy, supporting the automated detection of maize pests and diseases, and offering a robust solution for intelligent agricultural pest monitoring.

Accurate monitoring of the coverage and distribution of photosynthetic (PV) and non-photosynthetic vegetation (NPV) in the grasslands of semi-arid regions is crucial for understanding the environment and addressing climate change. However, the extraction of PV and NPV information from Unmanned Aerial Vehicle (UAV) remote sensing imagery is often hindered by challenges such as low extraction accuracy and blurred boundaries. To overcome these limitations, this study proposed an improved semantic segmentation model, designated SegFormer-CPED. The model was developed based on the SegFormer architecture, incorporating several synergistic optimizations. Specifically, a Convolutional Block Attention Module (CBAM) was integrated into the encoder to enhance early-stage feature perception, while a Polarized Self-Attention (PSA) module was embedded to strengthen contextual understanding and mitigate semantic loss. An Edge Contour Extraction Module (ECEM) was introduced to refine boundary details. Concurrently, the Dice Loss function was employed to replace the Cross-Entropy Loss, thereby more effectively addressing the class imbalance issue and significantly improving both the segmentation accuracy and boundary clarity of PV and NPV. To support model development, a high-quality PV and NPV segmentation dataset for Hengshan grassland was also constructed. Comprehensive experimental results demonstrated that the proposed SegFormer-CPED model achieved state-of-the-art performance, with a mIoU of 93.26% and an F1-score of 96.44%. It significantly outperformed classic architectures and surpassed all leading frameworks benchmarked here. Its high-fidelity maps can bridge field surveys and satellite remote sensing. Ablation studies verified the effectiveness of each improved module and its synergistic interplay. Moreover, this study successfully utilized SegFormer-CPED to perform fine-grained monitoring of the spatiotemporal dynamics of PV and NPV in the Hengshan grassland, confirming that the model-estimated f_PV and f_NPV were highly correlated with ground survey data. The proposed SegFormer-CPED model provides a robust and effective solution for the precise, semi-automated extraction of PV and NPV from high-resolution UAV imagery.

Photosynthetic vegetation (PV) and non-photosynthetic vegetation (NPV) play significant roles in ecosystem functions and ecological succession. Accurate monitoring of the coverage and distribution of PV and NPV in the grasslands of semi-arid regions is crucial for understanding the environment and addressing climate change. This study examined the Hengshan grassland site in China's semi-arid regions using imagery from unmanned aerial vehicles (UAVs), constructing a semantic segmentation label database via multiscale parameter optimisation, feature indicator selection, and manual correction methods. Three deep learning semantic segmentation models — PSPNet, DeepLabV3+, and U-Net—were employed to extract and compare the PV and NPV to determine the optimal semantic segmentation model. The experimental results showed that the PSPNet model exhibited a superior performance, with an overall classification accuracy of 89.2% and a Kappa coefficient of 0.80. These values were 0.8% and 3.9% higher and 0.02 and 0.07 higher than the corresponding values for DeepLabV3+ and U-Net, respectively. Further generalisability tests indicated that PSPNet achieved an overall classification accuracy of 87.5%–91.5% and a Kappa coefficient of 0.77–0.93 in different scenarios, effectively extracting the PV and NPV in various scenes of Hengshan grassland. Additionally, compared to estimates based on Sentinel-2A imagery, the UAV-based estimates of the fractional PV ( f _PV ) and fractional NPV ( f _NPV ) were closer to the results of field surveys. The method proposed in this study effectively extracted PV and NPV in China’s Hengshan grassland and demonstrated high reliability and applicability for long-term grassland monitoring. Therefore, the proposed approach can significantly contribute to the intelligent protection and sustainable management of grassland ecosystems in semi-arid areas.