Research on Apple Tree Leaf Disease Recognition Model Based on Multi Source Data Fusion and Deep Learning

In agriculture, precisely identifying apple leaf diseases is crucial for ensuring high yields. To tackle the opacity of neural networks, we propose a multi - source fusion framework that combines visible - light images and environmental sensors, thereby overcoming the limitations of single - modal data. The interpretable GAM - MFNet utilizes a generalized additive model. It processes structured data and image data through parallel modules, integrating univariate functions with convolutional networks. This design strikes a balance between prediction performance and global/local interpretability by means of feature analysis and prototype matching. Experiments demonstrate an accuracy of 89.99% in identifying five diseases, representing a 4.11% improvement over baselines. Feature visualization reveals that environmental factors such as temperature and humidity interact with lesion textures, highlighting the impacts of microclimate on disease symptoms. This research promotes agricultural diagnosis and provides a scalable and interpretable machine - learning framework for precision farming.