Hi

Im trying to build and train a Machine Learning model that autonomously perform color matching between the target gemstone and the Reference Standard color chart. A digital photo image of the target gemstone is firstly captured in a controlled environment in terms of illumination and background. This digital image is further pre-processed and fed into an algorithm that recognize and match its color distribution to the closest color in the Reference Standard color chart. Numerous Reference Standards exist but I will use the ColorCODEX (this link [ColorCODEX][1])

So I would like to know which Machine Learning Model to use in this case to ensure high matching accuracy and like what performance metric can I use to measure matching accuracy and the color space for the color model. And at the end what image per-processing need to be done?

[1]: https://static1.squarespace.com/static/5eb840daa2c9a8275e63081e/t/5ed13d0b02ae5147573d1e01/1590770964016/RP_2017_ColorCodex.pdf