Automated Image-based Identification of Forest Species: Challenges and Opportunities for 21st Century Xylotheques

The fast and accurate identification of forest species is fundamental to support their conservation, sustainable management, and, more specifically, the fight against illegal logging. Traditionally, identifications are done by using dichotomous or polytomous keys based on physical characteristics of trees. However, these techniques are of little use when the trees have been cut, removed from their natural environment, and consequently there is only a partial subset of information on all those traits. In these cases, it may be possible to resort to the anatomical characteristics of the wood, which are less affected by environmental factors and therefore have a high diagnostic value in the identification. For some years now, computers have been used to support the identification processes through interactive keys and access to global repositories of digital images, among others. However, techniques based on machine learning have recently been developed and applied successfully to the identification of both plant and animal species. Consequently, automatic or semiautomatic techniques have been proposed to support botanists, taxonomists and non-experts in the species identification process. This article presents an overview of the use of these techniques as well as the current challenges and opportunities for the identification of forest species based on xylotheque samples.