This study investigates the classification of pigment-manufacturing processes using deep learning to identify the optimal model for cultural property preservation science. Four convolutional neural networks (CNNs) (i.e., AlexNet, GoogLeNet, ResNet, and VGG) and one vision transformer (ViT) were compared on micrograph datasets of various pigments. Classification performance indicators, receiver-operating characteristic curves, precision–recall curves, and interpretability served as the primary evaluation measures. The CNNs achieved accuracies of 97–99%, while the ViT reached 100%, emerging as the best-performing model. These findings indicate that the ViT has potential for recognizing complex patterns and correctly processing data. However, interpretability using guided backpropagation approaches revealed limitations in the ViT ability to generate class activation maps, making it challenging to understand its internal behavior through this technique. Conversely, CNNs provided more detailed interpretations, offering valuable insights into the learned feature maps and hierarchical data processing. Despite its interpretability challenges, the ViT outperformed the CNNs across all evaluation metrics. This study underscores the potential of deep learning in classifying pigment manufacturing processes and contributes to cultural property conservation science by strengthening its scientific foundation for the conservation and restoration of historical artifacts.

One paper fragment and three pigments were excavated from earthenware found in the Hoeamsa Temple site (Historic Site No. 128), which is presumed to have been constructed in the 12th century. This study identifies the types and characteristics of the paper and pigments through methods such as fiber identification, pigment component analysis, and crystal structure analysis. Using the Graff 'C' staining method, the paper fragment is identified as paper-mulberry bast fiber (B. kazinoki Siebold), and according to radiocarbon dating results, the paper is believed to be from the Joseon Dynasty, between 1460 and 1640, which is when the Hoeamsa temple was believed to have been rebuilt. The green pigment consists of cerussite, malachite, and ataccamite, while quartz, muscovite, and illite were detected in the white pigment, and quartz, hematite, illite, and kaolinite were detected in the red pigment. The L^* value ranged from 41.59 to 79.05. The white pigment showed the highest brightness value at 79.05, while the green pigment was measured at 58.20 and the red pigment at 41.59. The a^* value ranged from -2.25 to 22.75: the a* value of the red pigment was the highest at 22.75, the green pigment was the lowest at -2.25. The b^* value ranged from 7.44 to 26.26: the b^*value of the red pigment was the highest at 26.26 and the white pigment was the lowest at 7.44.These scientific data identified the raw materials of paper used during the Joseon Dynasty. In addition, the data can be used for repair and restoration of ancient dancheong (traditional Korean decorative coloring on wooden buildings and artifacts) by analyzing the composition of pigments and identifying the characteristics of crystal structure.