Insects, representing one of the most species-rich biological groups, exert significant impacts on human society. Efficient and accurate insect identification plays a crucial role in agricultural and related domains. At present, deep learning approaches like convolutional neural networks (CNNs) and Swin Transformers have introduced novel solutions for this field. Although Swin Transformer and CNN architectures demonstrate excellent performance in general computer vision tasks, they still struggle to achieve satisfactory accuracy when dealing with subtle inter-species variations among insects in the same biological family. To address this challenge, an Edge Feature-Enhanced Swin Transformer (EFST) is proposed, which integrates multi-scale insect edge feature maps with raw input images and employs hierarchical Swin Transformer architectures for discriminative feature extraction. The EFST architecture integrates two key modules: edge feature extraction and classification. The former can provide three kinds of edge feature maps: blurred, coarse, and fine edge maps. These hierarchical edge representations enable precise differentiation of inter-species morphological variations within insect families. Classification is completed by using Swin Transformer. Experimental results demonstrate that EFST significantly enhances overall performance in insect image classification. Compared with the baseline Swin Transformer, EFST shows marked improvements across multiple metrics including accuracy, precision, recall, and F1-score on public datasets.