Plant Identification: Experts vs. Machines in the Era of Deep Learning

Deep Learning Techniques Challenge Flora Experts
  • Pierre BonnetEmail author
  • Hervé Goëau
  • Siang Thye Hang
  • Mario Lasseck
  • Milan Šulc
  • Valéry Malécot
  • Philippe Jauzein
  • Jean-Claude Melet
  • Christian You
  • Alexis Joly
Part of the Multimedia Systems and Applications book series (MMSA)


Automated identification of plants and animals have improved considerably in the last few years, in particular thanks to the recent advances in deep learning. The next big question is how far such automated systems are from the human expertise. Indeed, even the best experts are sometimes confused and/or disagree between each others when validating visual or audio observations of living organism. A picture or a sound actually contains only a partial information that is usually not sufficient to determine the right species with certainty. Quantifying this uncertainty and comparing it to the performance of automated systems is of high interest for both computer scientists and expert naturalists. This chapter reports an experimental study following this idea in the plant domain. In total, nine deep-learning systems implemented by three different research teams were evaluated with regard to nine expert botanists of the French flora. Therefore, we created a small set of plant observations that were identified in the field and revised by experts in order to have a near-perfect golden standard. The main outcome of this work is that the performance of state-of-the-art deep learning models is now close to the most advanced human expertise. This shows that automated plant identification systems are now mature enough for several routine tasks, and can offer very promising tools for autonomous ecological surveillance systems.



Most of the work conducted in this paper was funded by the Floris’Tic initiative, especially for the support of the organization of the PlantCLEF challenge. Milan Šulc was supported by CTU student grant SGS17/185/OHK3/3T/13. Valéry Malécot was supported by ANR ReVeRIES (ref: ANR-15-CE38-0004-01). Authors would like to thank the botanists who accepted to participate to this challenge : Benoit Bock (PhotoFlora), Nicolas Georges (Cerema), Arne Saatkamp (Aix Marseille Université, IMBE), François-Jean Rousselot, and Christophe Girod.


  1. 1.
    Bonnet, P., Joly, A., Goëau, H., Champ, J., Vignau, C., Molino, J. F., Barthélémy Daniel & Boujemaa, N. (2016). Plant identification: man vs. machine. Multimedia Tools and Applications, 75(3), 1647–1665.CrossRefGoogle Scholar
  2. 2.
    Sulc, M., & Matas, J. (2017). Learning with noisy and trusted labels for fine-grained plant recognition. Working Notes of CLEF, 2017.Google Scholar
  3. 3.
    Ludwig, A. R., Piorek, H., Kelch, A. H., Rex, D., Koitka, S., & Friedrich, C. M. (2017). Improving model performance for plant image classification with filtered noisy images. Working Notes of CLEF, 2017.Google Scholar
  4. 4.
    Hang, S. T., & Aono, M. (2017). Residual network with delayed max pooling for very large scale plant identification. Working Notes of CLEF, 2017.Google Scholar
  5. 5.
    Lasseck, M. (2017). Image-based plant species identification with deep convolutional neural networks. Working Notes of CLEF, 2017.Google Scholar
  6. 6.
    Atito, S., Yanikoglu, B., & Aptoula, E. Plant Identification with Large Number of Species: SabanciU-GebzeTU System in PlantCLEF 2017.Google Scholar
  7. 7.
    Lee, S. H., Chang, Y. L., & Chan, C. S. (2017). Lifeclef 2017 plant identification challenge: Classifying plants using generic-organ correlation features. Working Notes of CLEF, 2017.Google Scholar
  8. 8.
    Toma, A., Stefan, L. D., & Ionescu, B. (2017). Upb hes so@ plantclef 2017: Automatic plant image identification using transfer learning via convolutional neural networks. Working Notes of CLEF, 2017.Google Scholar
  9. 9.
    Szegedy, C., Liu, W., Jia, Y., Sermanet, P., Reed, S., Anguelov, D., …& Rabinovich, A. (2015). Going deeper with convolutions. In Proceedings of the IEEE conference on computer vision and pattern recognition (pp. 1–9).Google Scholar
  10. 10.
    Goëau, H., Bonnet, P., & Joly, A. (2017). Plant identification based on noisy web data: the amazing performance of deep learning (lifeclef 2017). CEUR Workshop Proceedings.Google Scholar
  11. 11.
    Krause, J., Sapp, B., Howard, A., Zhou, H., Toshev, A., Duerig, T., …& Fei-Fei, L. (2016, October). The unreasonable effectiveness of noisy data for fine-grained recognition. In European Conference on Computer Vision (pp. 301–320). Springer International Publishing.CrossRefGoogle Scholar
  12. 12.
    Goëau, H., Bonnet, P., & Joly, A. (2015). LifeCLEF Plant Identification Task 2015. CEUR Workshop Proceedings.Google Scholar
  13. 13.
    Goëau, H., Joly, A., Bonnet, P., Selmi, S., Molino, J. F., Barthélémy, D., & Boujemaa, N. (2014). Lifeclef plant identification task 2014. In CLEF2014 Working Notes. Working Notes for CLEF 2014 Conference, Sheffield, UK, September 15–18, 2014 (pp. 598–615). CEUR-WS.Google Scholar
  14. 14.
    Deng, J., Dong, W., Socher, R., Li, L. J., Li, K., & Fei-Fei, L. (2009, June). Imagenet: A large-scale hierarchical image database. In Computer Vision and Pattern Recognition, 2009. CVPR 2009. IEEE Conference on (pp. 248–255). IEEE.Google Scholar
  15. 15.
    Farnsworth, E. J., Chu, M., Kress, W. J., Neill, A. K., Best, J. H., Pickering, J., …& Ellison, A. M. (2013). Next-generation field guides. BioScience, 63(11), 891–899.Google Scholar
  16. 16.
    Bock B. (2014) Référentiel des trachéophytes de France métropolitaine réalisé dans le cadre d’une convention entre le Ministère chargé de l’Ecologie, le MNHN, la FCBN et Tela Botanica. Editeur Tela Botanica. Version 2.01 du 14 février 2014.Google Scholar
  17. 17.
    Gaston, K. J., & O’Neill, M. A. (2004). Automated species identification: why not?. Philosophical Transactions of the Royal Society of London B: Biological Sciences, 359(1444), 655–667.CrossRefGoogle Scholar
  18. 18.
    Goëau, H., Bonnet, P., & Joly, A. (2016). Plant identification in an open-world (lifeclef 2016). In Working Notes of CLEF 2016-Conference and Labs of the Evaluation forum, évora, Portugal, 5–8 September, 2016. (pp. 428–439).Google Scholar
  19. 19.
    Goëau, H., Bonnet, P., & Joly, A., Yahiaoui I., Barthelemy D., Boujemaa N., Molino J.-f. (2012). The ImageCLEF 2012 Plant Identification Task. CEUR Workshop Proceedings.Google Scholar
  20. 20.
    Goëau, H., Joly, A., Bonnet, P., Bakic, V., Barthélémy, D., Boujemaa, N., & Molino, J. F. (2013, October). The imageCLEF plant identification task 2013. In Proceedings of the 2nd ACM international workshop on Multimedia analysis for ecological data (pp. 23–28). ACM.Google Scholar
  21. 21.
    Goëau, H., Bonnet, P., Joly, A., Boujemaa, N., Barthelemy, D., Molino, J. F., …& Picard, M. (2011, September). The ImageCLEF 2011 plant images classi cation task. In ImageCLEF 2011.Google Scholar
  22. 22.
    He, K., Zhang, X., Ren, S., & Sun, J. (2016). Deep residual learning for image recognition. In Proceedings of the IEEE conference on computer vision and pattern recognition (pp. 770–778).Google Scholar
  23. 23.
    Ioffe, S., & Szegedy, C. (2015, June). Batch normalization: Accelerating deep network training by reducing internal covariate shift. In International Conference on Machine Learning (pp. 448–456).Google Scholar
  24. 24.
    Jauzein P. (1995). Flore des champs cultivés. Num.3912, Editions Quae.Google Scholar
  25. 25.
    Jauzein P., Nawrot O. (2013). Flore d’Ile-de-France: clés de détermination, taxonomie, statuts, Editions Quae.Google Scholar
  26. 26.
    Krizhevsky, A., Sutskever, I., & Hinton, G. E. (2012). Imagenet classification with deep convolutional neural networks. In Advances in neural information processing systems (pp. 1097–1105).Google Scholar
  27. 27.
    Saatkamp, A., Affre, L., Dutoit, T., & Poschlod, P. (2011). Germination traits explain soil seed persistence across species: the case of Mediterranean annual plants in cereal fields. Annals of botany, 107(3), 415–426.CrossRefGoogle Scholar
  28. 28.
    Simonyan, K., & Zisserman, A. (2014). Very deep convolutional networks for large-scale image recognition. arXiv preprint arXiv:1409.1556.Google Scholar
  29. 29.
    Szegedy, C., Ioffe, S., Vanhoucke, V., & Alemi, A. A. (2017). Inception-v4, Inception-ResNet and the Impact of Residual Connections on Learning. In AAAI (pp. 4278–4284).Google Scholar
  30. 30.
    Tison, J. M., Jauzein, P., Michaud, H., & Michaud, H. (2014). Flore de la France méditerranéenne continentale. Turriers: Naturalia publications.Google Scholar

Copyright information

© Springer International Publishing AG, part of Springer Nature 2018

Authors and Affiliations

  • Pierre Bonnet
    • 1
    • 2
    Email author
  • Hervé Goëau
    • 1
    • 2
  • Siang Thye Hang
    • 3
  • Mario Lasseck
    • 4
  • Milan Šulc
    • 5
  • Valéry Malécot
    • 6
    • 7
  • Philippe Jauzein
    • 8
  • Jean-Claude Melet
    • 9
  • Christian You
    • 10
  • Alexis Joly
    • 11
  1. 1.CIRAD, UMR AMAPMontpellierFrance
  2. 2.AMAP, Univ Montpellier, CIRAD, CNRS, INRA, IRDMontpellierFrance
  3. 3.Toyohashi University of TechnologyToyohashiJapan
  4. 4.Museum Fuer Naturkunde BerlinLeibniz Institute for Evolution and Biodiversity ScienceBerlinGermany
  5. 5.Czech Technical University in PraguePragueCzech Republic
  6. 6.IRHS, Agrocampus-OuestRennesFrance
  7. 7.INRA, Université d’AngersAngersFrance
  8. 8.AgroParisTech UFR Ecologie Adaptations InteractionsParisFrance
  9. 9.Independent BotanistNercillacFrance
  10. 10.Société Botanique Centre OuestNercillacFrance
  11. 11.Inria ZENITH TeamMontpellierFrance

Personalised recommendations