Michael P. Pound

This paper develops a multi-resolution encoder-decoder network for 3D root segmentation in CT images, achieving higher accuracy than existing methods.

RootNav 2.0 uses deep learning to automatically extract root architectures from plant images, achieving comparable accuracy to semi-automatic methods with significantly increased speed.

The paper introduces a dataset and a deep learning model for accurate wheat spike and spikelet localization and counting in plant phenotyping.