2015 | OriginalPaper | Buchkapitel
A Crop/Weed Field Image Dataset for the Evaluation of Computer Vision Based Precision Agriculture Tasks
verfasst von : Sebastian Haug, Jörn Ostermann
Erschienen in: Computer Vision - ECCV 2014 Workshops
Aktivieren Sie unsere intelligente Suche, um passende Fachinhalte oder Patente zu finden.
Wählen Sie Textabschnitte aus um mit Künstlicher Intelligenz passenden Patente zu finden. powered by
Markieren Sie Textabschnitte, um KI-gestützt weitere passende Inhalte zu finden. powered by
In this paper we propose a benchmark dataset for crop/weed discrimination, single plant phenotyping and other open computer vision tasks in precision agriculture. The dataset comprises 60 images with annotations and is available online (
http://github.com/cwfid
). All images were acquired with the autonomous field robot Bonirob in an organic carrot farm while the carrot plants were in early true leaf growth stage. Intra- and inter-row weeds were present, weed and crop were approximately of the same size and grew close together. For every dataset image we supply a ground truth vegetation segmentation mask and manual annotation of the plant type (crop vs. weed). We provide initial results for the phenotyping problem of crop/weed classification and propose evaluation methods to allow comparison of different approaches. By opening this dataset to the community we want to stimulate research in this area where the current lack of public datasets is one of the barriers for progress.