Skip to Main content Skip to Navigation
Master Thesis

Phénotypage par analyse d’images RGB du taux de couverture du blé et des adventices durant la phase préfloraison : prédiction de la nuisibilité des adventices sur la biomasse du blé grâce à un modèle écophysiologique simplifié

Abstract : As of today, professional remote sensing has successfully been used to locate weeds in cereal crops. The goal of our study was to test if an RGB captor from a regular camera would enable to follow the growth and evolution in space and time of wheat and weed plants in two micro-plots. Pictures were taken at four separate dates during the tillering. Using a classification algorithm, wheat and weed cover were calculated for each picture. We analysed this cover data using comparison of variance, to check for any heterogeneity in the populations and to assess the harm of weed plants. An Eco physiological model was used to quantify the impact of weed on wheat biomass predictions. Finally, maps of risk areas were plotted with an algorithm based on the various cover data. The cover data showed high spatial and temporal heterogeneity for both populations on both plots, but no single trend linking weed abundance to a reduction of wheat development could be drawn. The model we chose did not show any evidence of harm caused by weed plants to the wheat populations, probably because it did not fit close enough to the observations. Areas with more weed or showing slower wheat growth were flagged on maps of the plots to orient further agricultural interventions. The next step could be using drones to photograph the plots.
Document type :
Master Thesis
Complete list of metadatas

https://dumas.ccsd.cnrs.fr/dumas-03120020
Contributor : Agrocampus Ouest <>
Submitted on : Monday, January 25, 2021 - 11:31:13 AM
Last modification on : Sunday, February 21, 2021 - 3:22:29 AM

File

2020_Mignon_Victor_SIV.pdf
Files produced by the author(s)

Licence


Distributed under a Creative Commons Attribution - NonCommercial - NoDerivatives 4.0 International License

Identifiers

  • HAL Id : dumas-03120020, version 1

Collections

Citation

Victor Mignon. Phénotypage par analyse d’images RGB du taux de couverture du blé et des adventices durant la phase préfloraison : prédiction de la nuisibilité des adventices sur la biomasse du blé grâce à un modèle écophysiologique simplifié. Sciences du Vivant [q-bio]. 2020. ⟨dumas-03120020⟩

Share

Metrics

Record views

7

Files downloads

2