2020/09/28 by Fredrik Westling, Westling, Fredrik, Dr James Underwood +3
Agricultural and Biological Sciences · Environmental Science · #Computer Vision and Pattern Recognition (cs.CV) #FOS: Computer and information sciences #Forest ecology and management #Horticultural and Viticultural Research #Remote Sensing and LiDAR Applications #Robotics (cs.RO)
paper · pdf · doi:10.48550/arxiv.2009.13727
openalex publication_date 2020/09/28 · openalex created_date 2022/07/25 · openalex updated_date 2026/07/28
Digitisation of fruit trees using LiDAR enables analysis which can be used to\nbetter growing practices to improve yield. Sophisticated analysis requires\ngeometric and semantic understanding of the data, including the ability to\ndiscern individual trees as well as identifying leafy and structural matter.\nExtraction of this information should be rapid, as should data capture, so that\nentire orchards can be processed, but existing methods for classification and\nsegmentation rely on high-quality data or additional data sources like cameras.\nWe present a method for analysis of LiDAR data specifically for individual tree\nlocation, segmentation and matter classification, which can operate on\nlow-quality data captured by handheld or mobile LiDAR. Our methods for tree\nlocation and segmentation improved on existing methods with an F1 score of\n0.774 and a v-measure of 0.915 respectively, while trunk matter classification\nperformed poorly in absolute terms with an average F1 score of 0.490 on real\ndata, though consistently outperformed existing methods and displayed a\nsignificantly shorter runtime.\n