Data for: An automatic recognition system of Brazilian flora species based on textural features of macroscopic images of wood

Published: 28 June 2020| Version 2 | DOI: 10.17632/cc78ftcdf9.2
Deivison Souza,
Joielan Xipaia Santos,
Helena Cristina Vieira,
Tawani Lorena Naide,
Silvana Nisgoski,
Luiz S. Oliveira


This dataset is related to the research article entitled "A system of automatic recognition of species of Brazilian flora based on textural characteristics of macroscopic images of wood", by Deivison Venicio Souza, Joielan Xipaia Santos, Helena Cristina Vieira, Tawani Lorena Naide, Silvana Nisgoski and Luiz Eduardo S Oliveira published in the journal Wood Science and Technology ( Wood samples were from the collection of the Wood Anatomy and Quality Laboratory (LANAQM) of Federal University of Paraná (UFPR), located in Curitiba, Paraná. The wood samples’ transversal surfaces were sanded with a 120 sandpaper and macroscopic images of 46 species were taken with a Zeiss Discovery V 12 stereomicroscope, with size of 2080 × 1540 pixels and 10× magnification. The captured images have a resolution of 150 dpi. A total of 1,901 macroscopic images were obtained. Of the 46 species used in this study, 7 are on Brazil’s official list of endangered species. More specifically, the species Araucaria angustifolia and Ocotea porosa are classified in the category “Endangered”, Cedrela fissilis, Bertholletia excelsa, Mezilaurus itauba and Swietenia macrophylla are part of the “Vulnerable” group, and Euxylophora paraensis is considered “Critically Endangered”.



Universidade Federal do Para - Campus Altamira, Universidade Federal do Parana


Computer Vision, Machine Learning, Species Identification