Ceramics have held cultural significance for centuries, evolving into symbols of status and commodities. An AI system developed by researchers in Malaysia can classify Chinese ceramics and predict their value with up to 99% accuracy using deep learning and real auction data from institutions like Sotheby’s and Christie’s.

Powered by an NVIDIA GeForce RTX 3090 GPU, the AI system analyzes decorative motifs, shapes, and kiln-specific craftsmanship of Chinese ceramics. It aims to make cultural appraisal more objective, scalable, and accessible to a wider audience, especially for younger collectors and digital archive projects.

The system uses a typological classification system for ceramic vessel shapes, allowing detailed analysis and classification of various shapes such as bottles, jars, plates, bowls, and pots. It also pairs a detection model with an algorithm that learned market value from real-world auction results, providing new perspectives in artifact pricing and dating.

The AI system not only classifies Chinese ceramics but also analyzes intricate decorative patterns, annotating images based on major pattern categories like plant patterns, animal motifs, landscapes, and human figures. It aims to extend the conversation on cultural heritage by exploring AI for other forms of visual heritage, such as historical murals and Cantonese opera costumes.

Despite using consumer-grade gaming hardware, the AI system is changing the way artifacts are appraised, offering new perspectives on the value of centuries-old craftsmanship. The project highlights the intersection of technology and tradition in the ongoing debate on what makes something culturally valuable.

Read more at NVIDIA: A Gaming GPU Helps Crack the Code on a Thousand-Year Cultural Conversation