Publication detail

One-Class Learning Weed Plants Detection on Multispectral Images

KOLAŘÍK, M. JONÁK, M. PŘINOSIL, J. KRAJSA, O. BURGET, R. GAJDACZEK, T.

Original Title

One-Class Learning Weed Plants Detection on Multispectral Images

Type

conference paper

Language

English

Original Abstract

Modern precision agriculture methods focus on efficient crop care procedures with targeted chemical applications. As current computer vision algorithms allow us to distinguish different plant species, spot-spraying systems for precise herbicide application only on weed plants are being developed and used in praxis. Many such systems rely on deep learning algorithms trained on large datasets containing all possible plant species. While the crop plants keep the same visual characteristics all around the world, the species composition of weed plants can differ significantly, leading to lower detection accuracy for weed species that are not represented in the training dataset. In this work, we apply the PatchCore and PaDiM, two state-of-the-art anomaly detection algorithms, to a multispectral dataset of corn plant images in a one-class learning paradigm. The best performing algorithm achieved AUROC 94.2 despite the high visual heterogeneity and scarcity of the input data. Our results suggest it is possible to train weed-detection algorithms on a limited dataset in a one-class learning setting to transform the species classification into an anomaly detection problem.

Keywords

One-Class Learning; Anomaly detection; Precision agriculture; Multispectral imaging; Deep learning; Computer vision

Authors

KOLAŘÍK, M.; JONÁK, M.; PŘINOSIL, J.; KRAJSA, O.; BURGET, R.; GAJDACZEK, T.

Released

11. 10. 2022

Publisher

IEEE

Location

Valencia, Spain

ISBN

979-8-3503-9866-3

Book

2022 14th International Congress on Ultra Modern Telecommunications and Control Systems and Workshops (ICUMT)

ISBN

2157-023X

Periodical

International Congress on Ultra Modern Telecommunications and Control Systems and Workshops

Number

2157-023X

State

unknown

Pages from

76

Pages to

79

Pages count

4

URL

https://ieeexplore.ieee.org/document/9943391

BibTex

@inproceedings{BUT180850,
  author="Martin {Kolařík} and Martin {Jonák} and Jiří {Přinosil} and Ondřej {Krajsa} and Radim {Burget} and Tomáš {Gajdaczek}",
  title="One-Class Learning Weed Plants Detection on Multispectral Images",
  booktitle="2022 14th International Congress on Ultra Modern Telecommunications and Control Systems and Workshops (ICUMT)",
  year="2022",
  journal="International Congress on Ultra Modern Telecommunications and Control Systems and Workshops",
  number="2157-023X",
  pages="76--79",
  publisher="IEEE",
  address="Valencia, Spain",
  doi="10.1109/ICUMT57764.2022.9943391",
  isbn="979-8-3503-9866-3",
  issn="2157-023X",
  url="https://ieeexplore.ieee.org/document/9943391"
}