Research on Wheat Disease Recognition Algorithm based on Deep Learning

Abstract

In order to meet the demand for efficient detection and diagnosis of wheat diseases, this study proposes an improved YOLOv8-based object detection method. First, a dataset of 1,500 wheat disease images was collected and annotated. Subsequently, using the PyTorch framework, we modified and trained the YOLOv8 model by replacing the original Anchor-Free detection head with a dyhead-prune detection head, which incorporates attention mechanisms and pruning operations, and adopted Inner-Wise-GIoU as the loss function. Experimental results show that, compared to the baseline model, the improved model achieves a 1.5% increase in mAP50 while reducing the number of parameters by 2.5%, effectively enhancing both detection accuracy and operational efficiency. Finally, TensorRT was utilized for inference acceleration, enabling real-time disease detection. This research provides an efficient and feasible approach for the automatic recognition and timely prevention of wheat diseases, playing a crucial role in ensuring crop yield, reducing losses, and advancing the digital development of “Internet + Agriculture.”