<p>Chrysanthemum holds significant potential for various applications. The diversity of chrysanthemum varieties directly affects their medicinal qualities and economic value, therefore, accurate classification is essential. To address this issue, this paper proposes a chrysanthemum classification method based on multi-scale feature fusion and dual-view. Specifically, a dual-path network is designed to extract multi-scale semantic features from dual perspectives. Additionally, a multi-scale feature extraction network incorporating residual connections is employed to expand the receptive field and enhance feature representation. Subsequently, a multi-scale feature fusion module is introduced to effectively fuse features from the dual pathways, thereby improving both classification accuracy and network robustness. Finally, a deep learning-based classification model is trained for different chrysanthemum varieties. Experimental results demonstrate that the proposed method achieves an overall accuracy of 95.13% on the chrysanthemum dataset, outperforming existing methods and demonstrating superior stability and robustness.</p>