Machine-learning assisted determination of best acquisition protocols in variety testing

Hadhami Garbouge, Salma Samiei, Pejman Rasti, David Marc ROUSSEAU

Keywords: machine learning, acquisition protocol, variety testing

Abstract: In this communication, we propose an attempt of pipeline to determine best acquisition protocols automatically from ensemble of images acquired in different conditions. This is discussed to be specially useful in the field of variety testing where similar traits are measured in non standardized ways with help of images. Illustrations are given on distinct traits including detection of head wheat and sugar beet leaves. We discuss the methodological questions opened by this pilot proposal.