Data synthesis methods for semantic segmentation in agriculture: A Capsicum annuum dataset.

•This paper proposes a novel methodology to generate synthetic data in agriculture.•Based on empirical measurements, a 3D plant model was created.•Randomised plants were generated with this model and rendered in realistic scenes.•A synthetic dataset was created with a corresponding annotated empirical dataset.•Deep learning models were trained on the datasets and show performance gain.