3D grape bunch model reconstruction from 2D images

Yan San Woo, Zhuguang Li, Shun Tamura, Prawit Buayai, Hiromitsu Nishizaki, Koji Makino, Latifah Munirah Kamarudin, Xiaoyang Mao

Abstract: Highlights•Reconstructing a 3D model of a grape bunch with uniquely identified berries from videos captured of a real grape field.•New clustering and neural network-based for uniquely identifying berries in 3D point cloud and refined with video frames.•A 2D instance segmentation DNN preprocesses to exclude background from 3D point cloud generation.