Timely detection of pathogen spores is fundamental to ensuring early intervention and reducing the spread of corn diseases, like northern corn leaf blight, corn head smut, and corn rust. Traditional spore detection methods struggle to identify spore-level targets within complex backgrounds. To improve the recognition accuracy of various maize disease spores, this study introduced the YOLOv8s-SPM model by incorporating the space-to-depth and convolution (SPD-Conv) layers, the Partial Self-Attention (PSA) mechanism, and Minimum Point Distance Intersection over Union (MPDIoU) loss function. First, we combined SPD-Conv layers into the Backbone of the YOLOv8s to enhance recognition performance on small targets and low-resolution images. To improve computational efficiency, the PSA mechanism was incorporated within the Neck layer of the network. Finally, MPDIoU loss function was applied to refine the localization performance of bounding boxes. The results revealed that the YOLOv8s-SPM model achieved 98.9% accuracy on the mixed spore dataset. Relative to the baseline YOLOv8s, the YOLOv8s-SPM model yielded a 1.4% gain in accuracy. The improved model significantly improved spore detection accuracy and demonstrated superior performance in recognizing diverse spore types under complex background conditions. It met the demands for high-precision spore detection and filled a gap in intelligent spore recognition for maize, offering an effective starting point and practical path for future research in this field.

Bacterial blight of soybean (BBS), caused by Pseudomonas syringae pv. glycinea, is one of the most devastating diseases in soybean with significant yield losses ranging from 4% to 40%. The timely detection of BBS is the foundation for disease control. However, traditional identification methods are inefficient and rely heavily on expert knowledge. Existing automated approaches have not achieved high accuracy in natural environments and often require advanced equipment and extensive training, limiting their practicality and adaptability. To overcome these challenges, we propose LeafDPN, an improved Dual-Path Network model enhanced with Vision Transformer blocks in the forward propagation function and SE blocks in the ConvBNLayer. These enhancements improved the model’s accuracy, receptive field, and feature expression capabilities. Experiments conducted on a self-constructed dataset of 864 expert-labeled images across three disease types demonstrated that LeafDPN achieved a 98.96% identification accuracy and the shorted iteration time in just 24 epochs. It outperformed 14 baseline models like HRNet and EfficientNet in terms of accuracy, training efficiency, and resource consumption. In addition, the proposed LeafDPN model has the potential to be applied in the identification of other plant diseases based on available datasets.