Shui script is a unique writing system of the Shui ethnic group in China and contains abundant information on ethnic culture, earning it the reputation of being the“encyclopedia”of the Shui people.However, its digital preservation and intelligent recognition face significant challenges due to the scarcity of data resources, the complexity and non-standardization of handwritten character forms, and the fact that semantic interpretation relies heavily on a limited number of inheritors.To address these issues, this study constructs a Shui Script Character Database and a Shui Script Intelligent Recognition Platform.The database adopts a four-level architecture of book-section-sentence-character, integrating character images, phonetic annotations, and explanatory glosses, while providing multi-path retrieval functions to effectively aggregate and manage dispersed Shui script documents.The intelligent recognition platform is built on a two-stage deep learning framework consisting of DBNet for text detection and CRNN for text recognition.By introducing a Unicode Private Use Area encoding scheme and employing data augmentation strategies, the platform is optimized for the low-resource characteristics, long-tail distribution, and complex layouts of Shui script data, thereby achieving high-accuracy recognition of handwritten Shui script texts.Experimental results demonstrate that the proposed method significantly outperforms existing general-purpose OCR solutions in Shui script text detection and recognition tasks.This research provides an effective technical solution for the digital preservation of endangered minority scripts such as Shui script and offers important reference value for digital humanities research on related ancient documents.