Towards end-to-end rice row detection in paddy fields exploiting two-pathway instance segmentation

Zhiying Chen, Yinghu Cai, Yaxin Liu, Zhanpeng Liang, Hao Chen, Ruijun Ma, Long Qi

Abstract: Highlights•Casting the rice row detection problem as an instance segmentation problem.•A two-pathway end-to-end framework is customized for robust rice row detection.•The clustering loss is devised to parametrize the segmented rice row instances.•The proposed model exhibited superior performances versus other approaches.