• Aucun résultat trouvé

Added value of plant height and multi-temporal modelling to improve the estimation of wheat above-ground traits based on proxy-sensing RGB cameras

N/A
N/A
Protected

Academic year: 2021

Partager "Added value of plant height and multi-temporal modelling to improve the estimation of wheat above-ground traits based on proxy-sensing RGB cameras"

Copied!
25
0
0

Texte intégral

(1)

Added value of plant height and multi-temporal modelling to improve

the estimation of wheat traits based on proxy-sensing RGB cameras.

Dandrifosse S., Carlier A., Bustillo Vazquez E., Bouvry A.,

Leemans V., Dumont B. and Mercatoris B.

NSABS

2020

This study is funded by the Walloon Region

FRIA grant + project D31-1385 « PhenWheat »

(2)

Introduction

Methods

Results

Conclusions

2

Measure growth and health status of crops using RGB camera

Non destructive

Automatic

Objective

Affordable

(3)
(4)

Introduction

Methods

Results

Conclusions

4

1. Record height

Observed

zone

2. Use images at time t-1

Height

(5)

Introduction

Methods

Results

Conclusions

5

Goals

I. Quantify added value of

height information…

II. Quantify added value of

multitemporal modelling…

…to estimate key target traits :

LAI

Biomass

Nitrogen

(6)

Introduction

Methods

Results

Conclusions

6

Trial field

• Location : Lonzée (Gembloux)

• Variety : Safari (250 seeds/m²)

• Sowing : 17-10-2018

• Soil : loamy, moderate drainage

• 4 replications

Data acquisition at 3 dates

176

194

216

Days after

(7)

Introduction

Methods

Results

Conclusions

7

Image acquisition

Reference measurements

• Dry matter

• LAI

• Nitrogen

• BBCH growth stage

• 2 RGB cameras : JAI USB-GO5000

• Cameras spaced from 50 mm

• 1 m nadir above crop

(8)

Introduction

Methods

Results

Conclusions

8

RGB image

Segmented image

Background

Plants

Height map

(9)

Introduction

Methods

Results

Conclusions

9

LAI

Biomass

Nitrogen

PLS regression

10-k cross-validation

Height

Color

Cover

Time t-1

(10)

Introduction

Methods

Results

Conclusions

10

GOAL I.

Added value of height

Total biomass (t/ha) Leaf biomass (t/ha) Nitrogen (kg/ha) LAI (m²/m²)

(11)

Introduction

Methods

Results

Conclusions

11

GOAL II.

Added value of

images at t-1

Total biomass (t/ha) Leaf biomass (t/ha) Nitrogen (kg/ha) LAI (m²/m²)

194

216

(12)

Introduction

Methods

Results

Conclusions

12

194

176

(13)

Introduction

Methods

Results

Conclusions

13

Height : SIGNIFICANT added value

(14)

Thank you for your attention

Any questions ?

(15)

References

• Baresel, J. P., Rischbeck, P., Hu, Y., Kipp, S., Hu, Y., Barmeier, G., & Mistele, B. (2017). Use of a digital camera as alternative method for non-destructive detection of the leaf chlorophyll content and the nitrogen nutrition status in wheat. Computers and Electronics in Agriculture, 140, 25–33.

• Fernández, E., Gorchs, G., & Serrano, L. (2019). Use of consumer-grade cameras to assess wheat N status and grain yield. PLoS ONE, 14(2), 1–18.

• Jia, L., Chen, X., Li, M., Cui, Z., & Zhang, F. (2009). Comparsion of multispectral reflectance with digital color image in assessing the winter wheat nitrogen status. IFIP International Federation for Information Processing, 294, • Michez, A., Bauwens, S., Brostaux, Y., Hiel, M. P., Garré, S., Lejeune, P., & Dumont, B. (2018). How far can

consumer-grade UAV RGB imagery describe crop production? A 3D and multitemporal modeling approach applied to Zea mays. Remote Sensing, 10(11).

• Prey, L., von Bloh, M., & Schmidhalter, U. (2018). Evaluating RGB imaging and multispectral active and hyperspectral passive sensing for assessing early plant vigor in winter wheat. Sensors (Switzerland), 18(9).

• Tavakoli, H., & Gebbers, R. (2019). Assessing Nitrogen and water status of winter wheat using a digital camera.

(16)

Introduction

Methods

Results

Conclusions

16

Fertilization modalities

and associated 2019 grain

yields

(17)

Introduction

Methods

Results

Conclusions

17

RGB image

Segmented image

Height map

PREDICTORS CC : Canopy Cover ---VI : 9 Vegetation indices ; ;   H : 4 Height predictors • Mean • Median • Percentile 95 • Standard deviation

(18)

Introduction

Methods

Results

Conclusions

18

Segmentation method : Excess Red (ExR) threshold

(19)

Introduction

Methods

Results

Conclusions

19

(20)

Introduction

Methods

Results

Conclusions

20

(21)

Introduction

Methods

Results

Conclusions

21

(22)

Introduction

Methods

Results

Conclusions

22

(23)

Introduction

Methods

Results

Conclusions

23

(24)

Introduction

Methods

Results

Conclusions

24

Add BBCH stage

(25)

Introduction

Methods

Results

Conclusions

25

DM (t/ha) DM leaves (t/ha) N (kg/ha) LAI (m²/m²)

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0.61 0.77 0.75 0.82 0.9 0.78 0.54 0.64 0.16 0.31 0.32 0.39 0.14 0.37 0.48 0.55 0.19 0.42 0.51 0.58 CC Havg R G B R²

Références

Documents relatifs

Keywords: Partial least squares, principal components, shrinkage, model selec- tion, LAD regression, BACON algorithm, Krylov space, eigen space, precondi- tioning.. R´ esum´ e :

Sum and product of algebraic elements: prove that they are algebraic either by linear algebra or by means of the theorem of the elementary symmetric func- tions (see §

Global burden of disease (GBD) An estimate of health gaps (q.v.) for a comprehensive set of disease and injury causes, and for major risk factors, in the world populations using

After having presented the mathematical foundations of this class of method, along with two families of standard polynomials, Legendre and Chebyshev ones, two simple test problems

The objectives of the present study were to (1) assess the validity of Web-based self-reported weight, height, and resulting BMI compared with measured data in a subsample of

L’archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des

Abstract The maximum range of a time-of-flight camera is limited by the period- icity of the measured signal. Beyond a certain range, which is determined by the signal frequency,

This study carried out winter wheat biomass estimation based on a new biomass model through hyperspectral remote sensing, and it was verified by the field