Deep4mix : using deep learning to monitor vegetation species dynamics in mixed crops
 

• Starting date:  1st March 2022
• Scientific field: Deep learning for proxydetection
• Funding: #DigitAg
• Supervisors: Marie Weiss (Emmah, Inrae), Alexis Joly (Zenith, Inria)
• #DigitAg :  Axe 5 : Fouille de données, analyse de données, extraction de connaissances, Axe 2 : Innovations en agriculture numérique, Challenge 1 : Le challenge agroécologique, Challenge 2 : Le phénotypage rapide

Keywords: Agroecology, Instance segmentation, Semantic segmentation, Automated vegetation characterization

Abstract: The agroecological transition requires the development and the assessment of new multiperformant, resilient and sustainable agroecosystems. However, these systems are currently lacking high-throughput, non-destructive and objective observation tools. The data deficiency is even more critical because of the higher complexity of agroecosystems such as mixed crops as compared to single crop systems. High-throughput observation tools based on close range imagery therefore appear as essential for rapidly characterizing and hence better understanding these new agrosystems. However, if these tools have now reached a certain maturity for the monitoring of monospecific crops, their use in agroecology remains limited. This project aims to understand to what extent close-range imagery can be used for field monitoring of the dynamics of the proportion and structure of species in a crop mixture. The methodological approach involves a preliminary step of species identification within the canopy using deep learning models. A new database of annotated images will be therefore first created, and based on data acquired on multispecies crop canopies in the Remix project, but also on data acquired on single crop systems including weeds or not. The contribution of ancillary information (RGB images in two viewing directions, 3D LiDAR point cloud) as input to the deep models will be also investigated. Finally, once the species are identified, the project aims to estimate new traits such as the proportion of species, the corresponding leaf area or the overlapping area between these species, as well as to use the dynamics of these traits to identify key events such as the date of species cover.

Contact : dubos@inrae.fr

Social networks: LinkedIn -  ResearchGate

Communications & Papers:

Thesis manuscript (lien vers texte intégral en libre-accès) : http://www.theses.fr/s201366

NODeJ: an ImageJ plugin for 3D segmentation of nuclear objects, Tristan Dubos, Axel Poulet, Geoffrey Thomson, Emilie Péry, Frédéric Chausse, Christophe Tatout, Sophie Desset, Josien C. van Wolfswinkel, Yannick Jacob; Submitted: BMC Bioinformatics doi: 10.1101/2021.11.26.470128

Deep learning ­– promises for 3D nuclear imaging: a guide for biologists, Guillaume MOUGEOT, Tristan Dubos, Frederic Chausse, Emilie Pery, Katja Graumann Graumann, Christophe Tatout, David E Evans, and Sophie Desset, J Cell Sci. 2022 Apr 1;135(7):jcs258986. doi: 10.1242/jcs.258986

Automated 3D bio-imaging analysis of nuclear organization by NucleusJ 2.0,Tristan Dubos, Axel Poulet, Céline Gonthier-Gueret, Guillaume Mougeot, Emmanuel Vanrobays, Yanru Li, Sylvie Tutois, Emilie Pery, Frédéric Chausse, Aline V. Probst, Christophe Tatout & Sophie Desset,Nucleus, 11:1, 315-329, doi: 10.1080/19491034.2020.1845012

RT States: systematic annotation of the human genome using cell type-specific replication timing programs, Axel Poulet, Ben Li, Tristan Dubos, Juan Carlos Rivera-Mulia, David M Gilbert, Zhaohui S Qin, Bioinformatics, Volume 35, Issue 13, 1 July 2019, Pages 2167–2176,doi :doi.org/10.1093/bioinformatics/bty957

Genetic and epigenetic variation in 5S ribosomal RNA genes reveals genome dynamics in Arabidopsis thaliana, Simon L, Rabanal FA, Dubos T, Oliver C, Lauber D, Poulet A, Vogt A, Mandlbauer A, Le Goff S, Sommer A, Duborjal H,