top of page

EARBOX is a tool that allows the capture of normalized images and the extraction of phenotypic variables of interest.

DEEP LEARNING image analysis

Open source electronics

Available for maize and we are currently working to adapt it to wheat, barley, triticale and sunflower.

Explore how EARBOX system can help you

How does it work ?

kernels.png
kernels.png
yield components_final.png
yield components_final_focus.png

PARTOUT

kernels.png

Access to yield components

The EARBOX system is designed to automate and leverage phenotypic measurements of experimental stations.

Perform more accurate and refined analysis for:

  • Multi-site trials

  • Genotypic studies

  • Crop model calibration

  • Characterization of genotypes response to the environments

kernels.png

Data at ear and grains scale

The data are produced at 2 scales: the ear and the grains along the ear.


All data are measured or calculated for each of the 6 images of the ear captured by the system. The 6 images can be considered as repetitions, and averaged to obtain values at the ear scale, and indices (e.g. standard deviation, etc.) concerning their stability around the ear.


For grains, it is possible to make averages according to their "level" (cohort number) from the base of the ear. You can also process grain data according to their position (cm) in relation to the base of the ear.

ear data_ear.png

Move your mouse here

ear data_data.png
Ear Panelsv2.png
kernels.png

Accurate and reliable

Earbox is a tool developed by scientists for scientists.

 

Our development methodology shows our rigor and our requirement to offer research a useful tool that can feed tomorrow's reflections and experiments. In this logic, the methodologies of phenotypic data extraction are presented in the method article: Earbox, an open tool for high-throughput measurement of the spatial organization of maize ears and inference of novel traits. We scanned and trained the neural network on the widest possible phenotypic diversity, so that it would be able to segment maize kernels regardless of their appearance. We then developed an algorithm to extract in post-processing all the phenotypic data, which were all compared to manual data or via other machines on more than 800 ears.


This project was initiated in partnership with the INRAE of Montpellier (France) and more particularly the LEPSE (Laboratory of ecophysiology of plants under environmental stress) and the experimental station of Mauguio, with whom the EARBOX was defined, tested and validated.

kernels.png
kernels.png

Flexible and evolutive

Very easy to use, the system adapts to your identification methods (by spike or by plot). It reads most barcodes and QR codes thanks to its USB handheld, but you can also use a keyboard.


The code of the acquisition system is totally OPEN (on request), and you can modify or replace it, for your specific applications.


Designed for several species, we have started in 2021 a collaboration with the LARIS (Laboratoire Angevin de Recherche en Ingénierie des Systèmes) of the University of Angers and the French GEVES (Groupe d&#