Colour Calibration for Agricultural Imaging: A Pilot Study

Cameras are becoming central to agricultural automation, where they are used to measure and interpret an unstructured biological environment. Applications such as monitoring plant health and predicting yield depend on consistent image formation, yet images often contain variation. A key question is whether the observed variation is caused by genuine biological diversity or introduced by the camera pipeline. We present a pilot study of the colour variation of ripe strawberries imaged in the field. Images are captured with fruit still on the plant, deriving ripe fruit annotations from in-field observations of what was subsequently picked. We compare the chromaticity variation of ripe strawberries in images rendered with the camera's default processing pipeline and in images rendered after calibration from a colour checker placed in each image. The calibration reduces the observed variation suggesting that the cause is not solely biological and that in-field calibration can improve the consistency of colour measurements in agricultural imaging.