Diamond Auto Grading - Deep Learning
Shirtal DiaCam is developing an auto-grading system (color & clarity) for polished diamonds powered by deep learning artificial intelligence algorithms based on its 5 years of experience and visual data of hundreds of thousands photographed diamonds.
The scientific development part of the venture is done by Matrix AI Center, a world leading information technology, AI & Deep Learning company.
We now have the ability to identify a diamond’s color matching the GIA grading reports. The trial resulted in an impressive 90% success rate with regard to diamond’s color. These overwhelming, initial results were achieved without any additional efforts or research. It is expected that after minor adjustments and fine-tuning, the results of deep research of all to reach 95% accuracy, which influence the results.
Deep learning technology can produce results similar to or better than those of a trained gemologist using a 10x loupe. Deep learning is a form of algorithm-based machine learning that uses artificial neural networks to mimic the process by which the human brain identifies patterns and develops cognitive models.
How It Works
Artificial intelligence deep learning technology will produce a grading system that evaluates diamonds based on GIA certificate parameters. In the initial phase, the platform is using existing images of diamonds already in DiaCam360's database. These scans were produced in a sterile environment using the unique DiaCam360 automatic imaging system.
The platform is using deep learning algorithms to produce results similar to those that would be obtained by a human gemologist using a 10x loupe. For some parameters, the deep learning algorithms will even produce results more accurate than those of a human being. Over time, the system’s accuracy will improve as more images are added to the database.