This dataset contains annotations for Stem Emerging Points (SEP) in RGB and NIR image data recorded in the scope of the Sugar Beets 2016 Dataset.

Agricultural robots could be the key element on the way to affordable and sustainable agriculture. Based on monitoring the field plants individually, phenotyping, fertilising and pest control tasks could be applied precisely. Detecting the SEPs is an important perception task, first, to position weeding or fertilising tools at the plant's center, and second, as a way to finding natural landmarks in the field environment. This dataset contains annotations for ~2000 image sets with a broad variance of plant species and growth stages and allows developing and evaluating SEP detection approaches.

Overview

This dataset has been used to train convolutional neural networks for the task of SEP detection in RGB and NIR data. The approach is described in From Plants to Landmarks: Time-invariant Plant Localization that uses Deep Pose Regression in Agricultural Fields, which was presented at the Agri-Food Robotics Workshop at IROS 2017.

The dataset is based on image sequences from the Sugar Beet 2016 dataset. Each sequence consists of 100 to 300 image sets of 966x1296 px RGB and NIR images. The RGB image has been aligned with the NIR image. For each image set the SEP positions have been annotated by hand following the policy described in the paper.

@article{chebrolu2017ijrr,
title = {Agricultural robot dataset for plant classification, localization and mapping on sugar beet fields},
author = {Nived Chebrolu and Philipp Lottes and Alexander Schaefer and Wera Winterhalter and Wolfram Burgard and Cyrill Stachniss},
journal = {The International Journal of Robotics Research},
year = {2017}
doi = {10.1177/0278364917720510},
}

@inproceedings{kraemer17iros,
  author = {Florian Kraemer and Alexander Schaefer and Andreas Eitel and Johan Vertens and Wolfram Burgard},
  title = {From Plants to Landmarks: Time-invariant Plant Localization that uses Deep Pose Regression in Agricultural Fields},
  booktitle = {IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) Workshop, Agri-Food Robotics},
  year = {2017},
  url = {http://ais.informatik.uni-freiburg.de/publications/papers/kraemer17iros.pdf}
}
			

License Agreement

The Sugar Beets 2016 dataset is released under the Creative Commons license CC BY-SA 4.0. The Plant Centroids Dataset is merely an extension and variation of the original content. It is provided under the same CC BY-SA 4.0 license.

Download

The attached archive contains image series recorded in different rows and on different days. The folders are named corresponding to the Sugar Beets 2016 rosbags that the raw data was extracted from. Separate subfolders contain the RGB, NIR, and annotation data.