Method for Classifying Apple Leaf Diseases Based on Dual Attention and Multi-Scale Feature Extraction

Image datasets acquired from orchards are commonly characterized by intricate backgrounds and an imbalanced distribution of disease categories, resulting in suboptimal recognition outcomes when attempting to identify apple leaf diseases. In this regard, we propose a novel model, named RFCA ResNet, equipped with dual attention mechanism multi-scale feature extraction capacity, more effectively tackle these issues. The incorporated into ResNet is potent tool for mitigating the detrimental effects complex backdrops on outcomes. Additionally, utilizing class balance technique conjunction focal loss, adverse unbalanced dataset classification accuracy can be minimized. RFB module enables us expand receptive field achieve extraction, both which critical superior performance ResNet. Experimental results demonstrate that significantly outperforms standard CNN network exhibiting marked improvements 89.61%, 56.66%, 72.76%, 58.77% terms rate, precision recall F1 score, respectively. It better than other approaches, performs well generalization, has some theoretical relevance practical value.