The bivalve Pinctada margaritifera has the capacity to produce the most varied and colourful pearls in the world. Colour expression in the inner shell is under combined genetic and environmental control and is correlated with the colour of pearls produced when the same individual is used as a graft donor. One major limitation when studying colour phenotypes is grader subjectivity, which leads to inconsistent colour qualification and quantification. Through the use of HSV (Hue Saturation Value) colour space, we created an R package named ‘ImaginR’ to characterise inner shell colour variations in P. margaritifera. Using a machine-learning protocol with a training dataset, ImaginR was able to reassign individual oysters and pearls to predefined human-based phenotype categories. We then tested the package on samples obtained in an experiment testing the effects of donor conditioning depth on the colour of the donor inner shell and colour of the pearls harvested from recipients following grafting and 20 months of culture in situ. These analyses successfully detected donor shell colour modifications due to depth-related plasticity and the maintenance of these modifications through to the harvested pearls. Besides its potential interest for standardization in the pearl industry, this new method is relevant to other research projects using biological models.