A Novel Benchmark RGBD Dataset for Dormant Apple Trees and Its Application to Automatic Pruning

Abstract

Dormant pruning is a necessary procedure in the field of specialty crop production. In order to mitigate the need of huge labor, automation of this pruning process has become a topic of utmost importance in the field of horticulture. 3D modeling and reconstruction is a major step in such robotics precision agriculture. In this paper, we introduce a new public dataset which can be used for reconstructing dormant apple trees. Our dataset comprises of 9 different apple trees in both indoor and outdoor evironment. The images are collected using a portable Kinect2 sensor. To the best of our knowledge, this is the first publicly available dataset for the application like 3D modeling of dormant trees. We hope that the dataset will provide the entire research community working towards mechanizing dormant pruning a baseline benchmark for evaluating different 3D reconstruction and modeling algorithms.