A systematic method for determining colour descriptor states using image analysis is proposed using pili (Canarium ovatum) as a model. Kernel images of 52 pili accessions from the core collection of the Institute of Crop Science and National Plant Genetic Resources Laboratory, University of the Philippines Los Baños were captured using a calibrated VideometerLab 3 setup. Colour descriptor states were derived from the average International Commission on Illumination lightness (L*), green–red (a*) and blue–yellow (b*) colour component values. Cluster analysis and subsequent colour-parameter averaging per cluster were performed to produce representative colour values of descriptor states. The Euclidian distance (Delta E) of 3.5 was used to cut the cluster into readily distinguishable colour differences resulting to three descriptor states – light brown, brown and dark brown. Continuous colour variation of brown colour was observed indicating a possible quantitative nature of the trait. The use of delta E in elucidating the descriptor lists served as a gauge in successfully identifying distinguishable variations between colours. The method described can be applied to the elucidation of colour descriptor states of all parts of the plant of all crop species.