3D fully convolutional neural networks with intersection over union loss for crop mapping from multi-temporal satellite images

Sina Mohammadi, Mariana Belgiu, Alfred Stein

Research output: Contribution to conferencePaperpeer-review

8 Citations (Scopus)
31 Downloads (Pure)

Abstract

Information on cultivated crops is relevant for a large number of food security studies. Different scientific efforts are dedicated to generate this information from remote sensing images by means of machine learning methods. Unfortunately, these methods do not take account of the spatial-temporal relationships inherent in remote sensing images. In our paper, we explore the capability of a 3D Fully Convolutional Neural Network (FCN) to map crop types from multi-temporal images. In addition, we propose the Intersection Over Union (IOU) loss function for increasing the overlap between the predicted classes and ground reference data. The proposed method was applied to identify soybean and corn from a study area situated in the US corn belt using multi-temporal Landsat images. The study shows that our method outperforms related methods, obtaining a Kappa coefficient of 91.8%. We conclude that using the IOU loss function provides a superior choice to learn individual crop types.

Original languageEnglish
Pages5834-5837
Number of pages4
DOIs
Publication statusPublished - 2021
Event2021 IEEE International Geoscience and Remote Sensing Symposium, IGARSS 2021 - Brussels, Brussels, Belgium
Duration: 12 Jul 202116 Jul 2021
https://igarss2021.com

Conference

Conference2021 IEEE International Geoscience and Remote Sensing Symposium, IGARSS 2021
Abbreviated titleIGARSS 2021
Country/TerritoryBelgium
CityBrussels
Period12/07/2116/07/21
Internet address

Keywords

  • Crop mapping
  • deep learning
  • fully convolutional neural networks
  • time series
  • ITC-GREEN
  • 2024 OA procedure

Fingerprint

Dive into the research topics of '3D fully convolutional neural networks with intersection over union loss for crop mapping from multi-temporal satellite images'. Together they form a unique fingerprint.

Cite this