Breaking Boundaries: AI and Spectroscopic Technology Revolutionize Jade Identification, Accuracy Nears 98%
【Industry Observation】
On May 15, 2024, Dr. Li Ming, a professor at the Gemology Institute of China University of Geosciences (Wuhan) and a distinguished expert at the National Gemstone Testing Center (NGTC), unveiled a groundbreaking research achievement at the “2024 International Gemstone Technology Innovation Summit” hosted by “Global Jade Network”. The new paradigm for jade identification, combining artificial intelligence (AI) deep learning with multi-dimensional spectroscopic analysis, has achieved an astonishing accuracy rate of 97.8% in distinguishing common natural jades from imitations. This figure not only significantly surpasses the average accuracy of traditional visual identification (approximately 75%-85%) but also markedly improves the efficiency and reliability of existing instrumental identification methods.
In recent years, with the continuous prosperity of the jade market, sophisticated counterfeiting techniques have also advanced, posing significant challenges to consumers and the industry. According to statistics from the Gem & Jewelry Trade Association of China (GAC), the number of counterfeit jade cases investigated nationwide in 2023 increased by 15% year-on-year, involving illicit gains amounting to hundreds of millions of RMB. Traditional identification methods, such as empirical judgment, density testing, and refractive index measurement, often fall short when confronted with high-tech synthetic materials or enhanced treatments. For instance, distinguishing B+C grade jadeite (acid-treated, resin-filled, and dyed) is highly prone to misidentification by visual inspection. Even experienced experts might require the aid of large equipment like infrared spectroscopy for auxiliary judgment, which is also time-consuming.
Dr. Li Ming's team has developed a new system centered on building a vast jade spectroscopic database, encompassing tens of thousands of spectral data points for dozens of natural jades, including jadeite, Hetian jade, Xiuyan jade, Dushan jade, as well as their common treatments, dyeing, filling, and various imitations. By integrating multiple techniques such as Fourier Transform Infrared Spectroscopy (FTIR), Raman Spectroscopy, and Ultraviolet-Visible Absorption Spectroscopy (UV-Vis), the system can acquire 'fingerprint' information about the microscopic structure and chemical composition of jade. Subsequently, deep learning algorithms like Convolutional Neural Networks (CNN) are employed for feature extraction and pattern recognition of these high-dimensional spectral data. Dr. Li stated, "Our trained AI model can complete analysis and provide identification conclusions for unknown samples within seconds, greatly shortening the identification cycle and reducing labor costs."
This technology has undergone a six-month trial operation at the NGTC Guangzhou Laboratory. According to Zhang Hua, director of the laboratory, out of over 5000 submitted samples, the new system successfully identified 99.2% of natural A-grade jade and accurately distinguished 96.5% of B-grade, C-grade, and other imitations. Its performance was particularly outstanding in identifying high-quality synthetic jadeite (such as hydrothermal synthetic jadeite), effectively preventing misjudgments that might occur with traditional methods. Director Zhang commented, "This technology is expected to become the industry standard for future jade identification, providing stronger assurance for consumers."
However, the promotion of this new technology also faces challenges. Firstly, there's the issue of equipment cost. A complete set of high-precision spectrometers and an AI computing platform involves a significant investment, posing a barrier for small and medium-sized appraisal institutions. Secondly, the continuous updating and maintenance of the database are crucial. As new materials and counterfeiting techniques constantly emerge, new data needs to be collected and the AI model retrained. Dr. Li Ming admitted, "We are actively seeking collaboration with more international appraisal institutions and research institutes to jointly improve the global jade spectroscopic database and explore the development of more cost-effective portable identification devices."
Overall, the combination of AI and spectroscopic technology has brought revolutionary changes to the field of jade identification. It not only enhances the accuracy and efficiency of appraisal but also provides strong technical support for combating counterfeiting and maintaining market order. In the future, with the further maturation and popularization of this technology, we have reason to believe that the integrity of the jade market will significantly improve, and consumers will have greater confidence and assurance when purchasing jade products.