TongueBench Dataset

TongueBench Dataset is a quality-corrected tongue image dataset containing 2,694 tongue photographs for research in tongue diagnosis and medical artificial intelligence. The dataset was curated through a structured quality-control workflow to improve image consistency and release readiness. The current public release provides the image collection together with file-level metadata for archiving, indexing, and reuse. TongueBench Dataset is associated with the paper “TongueBench: Performance of Prompt-Engineered Large Multimodal Models in Traditional Chinese Medicine Tongue Diagnosis.” If you use this dataset, please cite the associated paper. This study was supported by National Key R&D Program of China (No.2025YFF0523900) and National Natural Science Foundation of China (No. 82274411).