It is very important to improve rapidity and correctness of analysis.
A powerful chemical imaging analysis system by AI technology has been developed by a research team led by Ri Ok I, a lecturer at the Faculty of Distance Education, and Kim Yong Ok, a section head at the Faculty of Chemical Engineering. They introduced deep learning technology, a branch of mechanical learning that has been accepted as an innovative one in the field of AI technology in recent years, thus making it labor- and cost-effective.
They first developed a method of extracting the main colour from a solution image taken by a camera and converting it into RGB values.
Then, they built a neuron network by deep learning technology. Learning data for the engine should be more than ten thousand sheets but construction of as many experimental data is not easy, so experimental data had to undergo processing for improving learning efficiency.
Finally, they designed it so that the RGB values are put into the engine and output values are processed to show analysis results.
On the basis of this, they developed an analysis system for Android-based smart phones or tablet PCs.
They applied the system to the analysis of Ni plating solution.
The result showed that analysis of plating solutions could be conducted promptly and easily any time with high accuracy without recourse to any kind of chemical reagents and analytical apparatus, and that it can be used by non-professionals too.
© 2021 KumChaek University of Technology