Authors

Abstract

Aiming at the problem of grape leaf variety identification with unbalanced number of samples in natural background, the AF-Swin Transformer model is proposed in this study.Firstly, Focal Loss is used to effectively tackle data imbalance in grape leaves.Secondly,the AdamW optimizer is selected to better control model complexity and improve generalization.The results show that the training accuracy of the AF-Swin Transformer model is 98.72%, which is 7.87 percentage points higher than that of the original Swin Transformer model.Precision and recall improved by 4.4 and 4.3 percentage points,respectively.This study provides a scientific basis for identifying grape leaf varieties with imbalanced sample data.

Abstract in Chinese

针对自然背景下样本数量不平衡的葡萄叶片品种识别问题，本研究提出了AF-Swin Transformer模型。首先，引入Focal Loss,有效应对葡萄叶片数据不平衡，其次，选用AdamW优化器，更好地控制模型复杂度并提高泛化能力。结果表明，AF-Swin Transformer模型的训练集准确率为98.72%，与原始Swin Transformer模型相比，训练准确率提高了7.87个百分点；精准率和召回率分别提高了4.4和4.3个百分点。本研究为样本数据不平衡的葡萄叶片品种识别提供了科学依据。