eCrop: A Novel Framework for Automatic Crop Damage Estimation in Smart Agriculture

Alakananda Mitra, Anshuman Singhal, Saraju P. Mohanty, Elias Kougianos, Chittaranjan Ray

Research output: Contribution to journalArticlepeer-review

2 Scopus citations

Abstract

Natural disasters impact agriculture. Farmers incur large losses due to crop damage. Climate/weather-driven natural events or disasters are happening often and are causing billions of dollars in losses. Crop insurance provides economic stability to the agricultural industry to make up for losses. A crop insurance claim is an extensive process and it takes time to process claims. In this paper, we propose a proof-of-concept of the novel crop damage estimation method, eCrop which is a part of our proposed agriculture cyber-physical system. eCrop is a grid-based method. We also present a novel crop damage detection method. It is the core of eCrop. It is a Convolutional Siamese Neural Network (CSNN) based model. A meta-learning approach has been taken to train the model. An accuracy of 92.86 % has been achieved. Our eCrop method can be adapted to agricultural insurance claim processing to automatically estimate the crop damage. It is scalable to any size of the cropland and any type of crop.

Original languageEnglish (US)
Article number319
JournalSN Computer Science
Volume3
Issue number4
DOIs
StatePublished - Jul 2022
Externally publishedYes

Keywords

  • Contrastive loss
  • Convolutional siamese neural network (CSNN)
  • Crop damage estimation
  • Haversine formula
  • Insurance loss claim
  • Internet-of-agro-things (IoAT)
  • Small dataset
  • Smart agriculture

ASJC Scopus subject areas

  • Computational Theory and Mathematics
  • Computer Networks and Communications
  • Computer Science Applications
  • Computer Science(all)
  • Artificial Intelligence
  • Computer Graphics and Computer-Aided Design

Fingerprint

Dive into the research topics of 'eCrop: A Novel Framework for Automatic Crop Damage Estimation in Smart Agriculture'. Together they form a unique fingerprint.

Cite this