Staff directory

Nils Bourland

Biology
Wood biology

SmartwoodID

Smart classification of Congolese timbers: deep learning techniques for enforcing forest conservation
A substantial part of the timber traded each year is still illegal. Illegal logging is the most profitable biodiversity crime. It involves a high risk of irreversible damage to forests since it often implicates overexploitation of highly sought after, sometimes protected, species. This is especially pertinent for tropical species, as it is estimated that 30-90% of the tropical timber volume is harvested illegally (Deklerck et al., 2020; Hirschberger, 2008; Hoare, 2015; Vlam et al., 2018). Timber regulations are already active (CITES, FLEGT, EUTR , Amendment to the U.S. Lacey Act), but implementation and enforcement are a challenge. Wood identification is crucial in the enforcement process when it comes to verify whether the shipment corresponds with the products mentioned on the accompanying documents. For this reason, there is a growing demand for timber identification tools that can be applied by law enforcement officers. SmartwoodID aims at improving both identification success and speed by non-experts. The project aims at automating part of the wood identification process by applying artificial intelligence techniques for the analysis of wood anatomical images of timber species of the Democratic Republic of the Congo. The project focusses on 1190 Central African Timbers to create a database with high-resolution scans of the endgrain surface along with expert wood anatomical descriptions. The study material comes from the Tervuren Xylarium. This because said database offers the most complete collection of reference material for the development of wood classification and identification approaches for Congolese species, comprising more than 2000 woody species from the DRC (timber trees, small trees, shrubs, dwarf shrubs and lianas). The resulting database is used to make an illustrated key for wood identification. The project also takes advantage of the power of modern deep learning approaches. The scans and anatomical descriptions will therefore serve as annotated training data to develop a machine learning assisted illustrated key for wood identification.

A substantial part of the timber traded each year is still illegal. Logging illegally is the most profitable biodiversity crime. UN Environment estimates that illegal logging and the associated timber trade counts up from 50 to 152 billion USD per year. Illegal logging involves a high risk of irreversible damage to forests since it often involves overexploitation of highly sought after, sometimes protected, species. Timber regulations are already active (CITES, FLEGT, EUTR), but implementation and enforcement are a challenge. Currently, Belgium has the negative connotation of being the ‘hub of illegal timber trade’. 27.5% of the total imports of primary tropical timber products into the 28 countries of the European Union are imported via Belgium (mainly via the port of Antwerp). Wood identification is crucial in the enforcement process when it comes to verify whether the shipment corresponds with the products mentioned on the accompanying documents. For this reason, there is a growing demand for timber identification tools that can be applied by law enforcement officers. SmartwoodID aims at improving both identification success and speed by non-experts. The wood identification process will be automated by applying artificial intelligence techniques for the analysis of wood anatomical images of timber species of the Democratic Republic of the Congo. The tree flora of Central Africa comprises 3013 species, 27 of these belong to the class 1 commercial timber species of the DRC and are actually intensively logged and traded, 20 to class 2 (have potentially a big commercial value), 44 to class 3 (are considered to be promoted) and 879 to class 4 (commercial value is not yet known). The project uses xylarium specimens of all the species of the four classes and takes advantage of the power of modern deep learning approaches. It also relies on expert wood anatomical descriptions which will serve as annotated training data to develop the software.

Principal investigator:

  • Hans Beeckman
  • Dates:

    2021 2024

    External collaborators:

    Prof. Jan Van den Bulcke (Ghent University)
    Prof. Jan Verwaeren (Ghent University)
    Prof. Tom De Mil (Université de Liège-campus Gembloux)