A Dataset and Benchmark for Traditional Chinese Official Hat Morphology Analysis and AI Recognition

READMEProject TitleA Dataset and Benchmark for Traditional Chinese Official Hat Morphology Analysis and AI RecognitionVersion1.0Dataset & Code DescriptionThis repository provides the dataset, preprocessing pipeline, training code, and visualization scripts for research on Chinese official hat (Guanmao) culture, morphological classification, and design cognition simulation.Materials Included Data from OMS: Raw image & map data of official hats Cleaned Data: Preprocessed, cropped, and categorized image data Caps data: Hat morphology–labeled dataset label data: Classification labels for hat types, dynasties, identities, and features metadata: Structured metadata (file name, identity, dynasty, hat description, symmetry, wings, source, remarks) guanmao_research_main.py: Model training script test_and_visualize.py: Model evaluation and result visualization trans_into_picture.py: Raw data extraction and image conversion Usage Pipeline Run trans_into_picture.py to extract raw map data from OMS. Clean, crop, and categorize raw data to generate Cleaned Data. Run guanmao_research_main.py to train the model and obtain the best-performing checkpoint. Run test_and_visualize.py to evaluate the model on the test set and visualize prediction results. Data AvailabilityAll data and code in this repository are available for academic use. If you use this dataset in your research, please cite this repository and the corresponding paper.