PhenoBench

Dense Crop and Weed annotations

Semantics for crops (sugar beet) and weeds.

Plant instances

Accurate plant instance annotations of crops and weeds.

Leaf instances

Leaf-level annotations of crop leaves.

High resolution images

100 mega pixel UAV images of real agricultural fields.

Plant and Leaf Annotations

Dense pixel-wise annotations of over 5,000 plants including crops and weeds. Highly detailed, pixel-wise annotations of over 30,000 crop leaves.

Different field conditions

We captured data at multiple days in two different years resulting in different growth stages and lighting conditions.

Benchmarks

Benchmarks for semantic segmentation, panoptic segmentation, detection, leaf instance segmentation, and hierarchical panoptic segmentation.

News

December 03, 2032 — PhenoBench (Version 1.1.0): Temporally consistent instance ids that identify crops by a unique instance id. Images and semantics are the same. See the dataset webpage for a download link.

July 15, 2023 — Published baseline code, checkpoints, and predictions of all baselines on GitHub.

July 13, 2023 — First release of our PhenoBench devkit on GitHub and PyPi.

June 21, 2023 — Benchmark tasks are online! See our benchmarks page for more information and links to CodaLab competitions.

June 20, 2023 — Start of PhenoBench Competition at CVPPA 2023 Workshop at ICCV 2023!

June 01, 2023 — Dataset release (Version 1.0.0). See the dataset webpage for a download link.

Examples

(Please hover over images to have a closer look.)

Plant Semantics

Plant Instances

Crop Leaf Instances

Team

Jan Weyler

Federico Magistri

Elias Marks

Yue Linn Chong

Matteo Sodano

Gianmarco Roggiolani

Nived Chebrolu

Cyrill Stachniss

Jens Behley

Citation


@article{weyler2023dataset,
  author = {Jan Weyler and Federico Magistri and Elias Marks and Yue Linn Chong and Matteo Sodano 
    and Gianmarco Roggiolani and Nived Chebrolu and Cyrill Stachniss and Jens Behley},
  title = {{PhenoBench --- A Large Dataset and Benchmarks for Semantic Image Interpretation
    in the Agricultural Domain}},
  journal = {arXiv preprint},
  year = {2023}
}

Acknowledgements

This work has been funded by been funded by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) under Germany's Excellence Strategy, EXC-2070 - 390732324 (PhenoRob).