2020/07/17 by Jordan Ubbens, Ubbens, Jordan, Tewodros Ayalew +9
Agricultural and Biological Sciences · Biochemistry, Genetics and Molecular Biology · Environmental Science · #Computer Vision and Pattern Recognition (cs.CV) #FOS: Computer and information sciences #Plant Pathogens and Fungal Diseases #Remote Sensing in Agriculture #Smart Agriculture and AI
paper · pdf · doi:10.48550/arxiv.2007.09178
openalex publication_date 2020/07/17 · openalex created_date 2022/07/26 · openalex updated_date 2026/07/28
Counting plant organs such as heads or tassels from outdoor imagery is a\npopular benchmark computer vision task in plant phenotyping, which has been\npreviously investigated in the literature using state-of-the-art supervised\ndeep learning techniques. However, the annotation of organs in field images is\ntime-consuming and prone to errors. In this paper, we propose a fully\nunsupervised technique for counting dense objects such as plant organs. We use\na convolutional network-based unsupervised segmentation method followed by two\npost-hoc optimization steps. The proposed technique is shown to provide\ncompetitive counting performance on a range of organ counting tasks in sorghum\n(S. bicolor) and wheat (T. aestivum) with no dataset-dependent tuning or\nmodifications.\n