Deep learning for weed identification based on seed images

Francisco Pando; Ignacio Heredia; Carlos Aedo Pérez; Mauricio Velayos Rodríguez; Lara Lloret Iglesias; Joel Calvo

doi:10.3897/biss.2.25749

Biodiversity Information Science and Standards : Conference Abstract

Conference Abstract

Deep learning for weed identification based on seed images

Francisco Pando^‡, Ignacio Heredia^§, Carlos Aedo Pérez^‡, Mauricio Velayos Rodríguez^‡, Lara Lloret Iglesias^§, Joel Calvo^|

‡ Real Jardín Botánico - CSIC, Madrid, Spain

§ Insituto de Fisica de Cantabria - CSIC/UC, Santander, Spain

| Pontificia Universidad Católica de Valparaíso, Valparaíso, Chile

Corresponding author: Francisco Pando (pando@rjb.csic.es)

Received: 12 Apr 2018 | Published: 21 May 2018

This is an open access article distributed under the terms of the Creative Commons Attribution License (CC BY 4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Citation: Pando F, Heredia I, Aedo Pérez C, Velayos Rodríguez M, Lloret Iglesias L, Calvo J (2018) Deep learning for weed identification based on seed images. Biodiversity Information Science and Standards 2: e25749. https://doi.org/10.3897/biss.2.25749

Abstract

Reliable plant species identification from seeds is intrinsically difficult due to the scarcity of features and because it requires specialized expertise that is becoming increasingly rarer, as the number of field plant taxonomists is diminishing (Bacher 2012, Haas and Häuser 2005). On the other hand, seed identification is relevant in some science domains such as plant community ecology, archaeology, paleoclimatology. Besides, economic activities such as agriculture, require seed identification to assess weed species contained in the "soil seed banks" (Colbach 2014) to enable targeted treatments before they become a problem.

In this work, we explore and evaluate several approaches by using different training image sets with various requisites and assessing their performance with test datasets of different sources.

The core training dataset is provided by the Anthos project (Castroviejo et al. 2017) as a subset of its image collection. It consists of nearly a 1000 images of seeds identified by experts.

As identification algorithm, we will use state-of-the-art convolutional neural networks for image classification (He et al. 2016). The framework is fully written in Python using the TensorFlow (Abadi et al. 2016) module for deep learning.

Presenting author

Francisco Pando

Acknowledgements

Funding program

EU Horizon 2020 framework programme project DEEP-Hybrid-Datacloud (Grant Agreement number 777435)

Grant title

Hosting institution

Ethics and security

Author contributions

Conflicts of interest

References

Abadi M, Barham P, Chen J, Chen Z, Davis A, Dean J, al. e (2016)

TensorFlow: A System for Large-Scale Machine Learning

OSDI

265

‑

283

Bacher S (2012)

Still not enough taxonomists: reply to Joppa et al.

Trends in Ecology & Evolution

(

‑

. https://doi.org/10.1016/j.tree.2011.11.003

Castroviejo S, Aedo C, Medina L (2017)

Anthos. Sistema de Información de las Plantas de España (1999-2017). Real Jardín Botánico-CSIC / Fundación Biodiversidad

. www.anthos.es. Accessed on: 2018-2-20.

Colbach N (2014)

The functional role of the soil seed bank in agricultural ecosystems.

Gallagher, R. S.(ed.). Seeds: the ecology of regeneration in plant communities

3erd

CABI

Wallingford, UK

. https://doi.org/10.1079/9781780641836.0235

Haas F, Häuser CL (2005)

Taxonomists: an endangered species?

(Secretariat of the Convention on Biological Diversity, ed.), Technical Series

‑

He K, Zhang X, Ren S, Sun J (2016)

Deep residual learning for image recognition

2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR)

(pp. 770-778)

pp. https://doi.org/10.1109/cvpr.2016.90

Supplementary material

Endnotes