As one of the six classical scripts in the world, the Yi script holds immense cultural significance, and its recognition plays a pivotal role in preserving and digitizing the cultural heritage of ethnic minorities. This study presents the first dedicated Yi character sample library for the Daliangshan region of Sichuan Province, addressing the challenges posed by limited data availability and inadequate algorithmic adaptability in Yi script recognition. Moreover, we propose a lightweight recognition model built on an enhanced MobileNetV3 architecture. By optimizing the network’s depth and width, the model significantly boosts feature extraction capabilities while maintaining a compact structure. Experimental results reveal that the model achieves a recognition accuracy of 94.6% on Yi character datasets. This study provides an effective technical solution for digitizing ancient literature and enabling the intelligent processing of ethnic languages and scripts, offering valuable practical insights for advancing informatization in minority regions.