Detection of crop diseases using enhanced variability imagery data and convolutional neural networks

Shai Kendler, Ran Aharoni, Sierra Young, Hanan Sela, Tamar Kis-Papo, Tzion Fahima, Barak Fishbain

Research output: Contribution to journalArticlepeer-review

Abstract

The timely detection of crop diseases is critical for securing crop productivity, lowering production costs, and minimizing agrochemical use. This study presents a crop disease identification method that is based on Convolutional Neural Networks (CNN) trained on images taken with consumer-grade cameras. Specifically, this study addresses the early detection of wheat yellow rust, stem rust, powdery mildew, potato late blight, and wild barley net blotch. To facilitate this, pictures were taken in situ without modifying the scene, the background, or controlling the illumination. Each image was then split into several patches, thus retaining the original spatial resolution of the image while allowing for data variability. The resulting dataset was highly diverse since the disease manifestation, imaging geometry, and illumination varied from patch to patch. This diverse dataset was used to train various CNN architectures to find the best match. The resulting classification accuracy was 95.4 ± 0.4%. These promising results lay the groundwork for autonomous early detection of plant diseases. Guidelines for implementing this approach in realistic conditions are also discussed.
Original languageEnglish
Article number106732
Number of pages8
JournalComputers and Electronics in Agriculture
Volume193
DOIs
StatePublished - Feb 2022

Bibliographical note

Publisher Copyright:
© 2022 Elsevier B.V.

Keywords

  • Crop disease
  • Convolutional Neural Networks
  • Classification
  • Precision agriculture
  • Generalization

ASJC Scopus subject areas

  • Forestry
  • Agronomy and Crop Science
  • Computer Science Applications
  • Horticulture

Fingerprint

Dive into the research topics of 'Detection of crop diseases using enhanced variability imagery data and convolutional neural networks'. Together they form a unique fingerprint.

Cite this