---
dataset_info:
  features:
  - name: image
    dtype: image
  - name: label
    dtype: image
  splits:
  - name: train
    num_bytes: 71849508.0
    num_examples: 48
  - name: val
    num_bytes: 17901648.0
    num_examples: 12
  download_size: 89729067
  dataset_size: 89751156.0
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*
  - split: val
    path: data/val-*
---
A Crop/Weed Field Image Dataset
=============

The Crop/Weed Field Image Dataset (CWFID) accompanies the following publication: "Sebastian Haug, Jörn Ostermann: A Crop/Weed Field Image Dataset for the Evaluation of Computer Vision Based Precision Agriculture Tasks, [CVPPP 2014](http://www.plant-phenotyping.org/CVPPP2014) Workshop, [ECCV 2014](http://eccv2014.org)"

This dataset comprises field [images](images), vegetation segmentation [masks](masks) and crop/weed plant type [annotations](annotations). The paper provides details, e.g. on the field setting, acquisition conditions, image and ground truth data format.

Paper
-----
Paper available [here](http://rd.springer.com/chapter/10.1007%2F978-3-319-16220-1_8).

Bibtex:
```
@inproceedings{haug15,
  author={Haug, Sebastian and Ostermann, J{\"o}rn},
  title={A Crop/Weed Field Image Dataset for the Evaluation of Computer Vision Based Precision Agriculture Tasks},
  year={2015},
  booktitle={Computer Vision - ECCV 2014 Workshops},
  doi={10.1007/978-3-319-16220-1_8},
  url={http://dx.doi.org/10.1007/978-3-319-16220-1_8},
  pages={105--116},
}
```

Use
---
All data is subject to copyright and may only be used for non-commercial research. In case of use please cite our publication.

Contact Sebastian Haug (sebastian.haug@de.bosch.com) for any questions.