This study proposes a novel Transformer-based multimodal fusion framework for grape disease detection, integrating RGB images, hyperspectral data, and environmental sensor readings. Unlike traditional single-modal approaches, the proposed method leverages a Transformer-based architecture to effectively capture spatial, spectral, and environmental dependencies, improving disease detection accuracy under varying conditions. A comprehensive dataset was collected, incorporating diverse lighting, humidity, and temperature conditions, and enabling robust performance evaluation. Experimental results demonstrate that the proposed method outperforms state-of-the-art (SOTA) models, achieving an mAP@50 of 0.94, an mAP@75 of 0.93, Precision of 0.93, and Recall of 0.95, surpassing leading detection baselines. The results confirm that the integration of multimodal information significantly enhances disease detection robustness and generalization, offering a promising solution for real-world vineyard disease management.

The rapid advancement in smart agriculture has introduced significant challenges, including data scarcity, complex and diverse disease features, and substantial background interference in agricultural scenarios. To address these challenges, a disease detection method based on few-shot learning and diffusion generative models is proposed. By integrating the high-quality feature generation capabilities of diffusion models with the feature extraction advantages of few-shot learning, an end-to-end framework for disease detection has been constructed. The experimental results demonstrate that the proposed method achieves outstanding performance in disease detection tasks. Across comprehensive experiments, the model achieved scores of 0.94, 0.92, 0.93, and 0.92 in precision, recall, accuracy, and mean average precision (mAP@75), respectively, significantly outperforming other comparative models. Furthermore, the incorporation of attention mechanisms effectively enhanced the quality of disease feature representations and improved the model’s ability to capture fine-grained features.

Accurate detection of soybean diseases is a critical component in achieving intelligent agricultural management. However, traditional methods often underperform in complex field scenarios. This paper proposes a diffusion-based object detection model that integrates the endogenous diffusion sub-network and the endogenous diffusion loss function to progressively optimize feature distributions, significantly enhancing detection performance for complex backgrounds and diverse disease regions. Experimental results demonstrate that the proposed method outperforms multiple baseline models, achieving a precision of 94%, recall of 90%, accuracy of 92%, and mAP@50 and mAP@75 of 92% and 91%, respectively, surpassing RetinaNet, DETR, YOLOv10, and DETR v2. In fine-grained disease detection, the model performs best on rust detection, with a precision of 96% and a recall of 93%. For more complex diseases such as bacterial blight and Fusarium head blight, precision and mAP exceed 90%. Compared to self-attention and CBAM, the proposed endogenous diffusion attention mechanism further improves feature extraction accuracy and robustness. This method demonstrates significant advantages in both theoretical innovation and practical application, providing critical technological support for intelligent soybean disease detection.