Feature Extraction and Machine Learning for the Classification of Brazilian Savannah Pollen Grains

Ariadne Barbosa Gonçalves; Junior Silva Souza; Gercina Gonçalves da Silva; Marney Pascoli Cereda; Arnildo Pott; Marco Hiroshi Naka; Hemerson Pistori

doi:10.1371/journal.pone.0157044

Feature Extraction and Machine Learning for the Classification of Brazilian Savannah Pollen Grains

PLoS One. 2016 Jun 8;11(6):e0157044. doi: 10.1371/journal.pone.0157044. eCollection 2016.

Authors

Ariadne Barbosa Gonçalves¹, Junior Silva Souza², Gercina Gonçalves da Silva³, Marney Pascoli Cereda³, Arnildo Pott⁴, Marco Hiroshi Naka^{1

5}, Hemerson Pistori^{1

2

3}

Affiliations

¹ Department of Biotechnology, INOVISAO, Dom Bosco Catholic University, Campo Grande, Mato Grosso do Sul, Brazil.
² Department of Computing Science, Universidade Federal de Mato Grosso do Sul, Campo Grande, Mato Grosso do Sul, Brazil.
³ Department of Environmental Science and Agricultural Sustainability, Dom Bosco Catholic University, Campo Grande, Mato Grosso do Sul, Brazil.
⁴ Laboratory of Botany, Universidade Federal de Mato Grosso do Sul, Campo Grande, Mato Grosso do Sul, Brazil.
⁵ Direction of Research, Extension and Institutional, Federal Institute of Mato Grosso do Sul, Science and Technology. Campo Grande, Mato Grosso do Sul, Brazil.

Abstract

The classification of pollen species and types is an important task in many areas like forensic palynology, archaeological palynology and melissopalynology. This paper presents the first annotated image dataset for the Brazilian Savannah pollen types that can be used to train and test computer vision based automatic pollen classifiers. A first baseline human and computer performance for this dataset has been established using 805 pollen images of 23 pollen types. In order to access the computer performance, a combination of three feature extractors and four machine learning techniques has been implemented, fine tuned and tested. The results of these tests are also presented in this paper.

MeSH terms

Brazil
Grassland*
Machine Learning*
Pollen / anatomy & histology*
Pollen / classification*

Grants and funding

This work has received financial support from the Dom Bosco Catholic University, UCDB and the Foundation for the Support and Development of Education, Science and Technology from the State of Mato Grosso do Sul, FUNDECT. Some of the authors have been awarded with Scholarships from the the Brazilian National Council of Technological and Scientific Development, CNPq and the Coordination for the Improvement of Higher Education Personnel, CAPES.