This project aims at digitizing ~6% of the species (~5,000 wood specimens) of the vast Tervuren xylarium, building an illustrated key for timber identification and developing a machine learning ‘assistant’ that will be easily accessible to law enforcement officers. The Tervuren xylarium serves since long as reference collection for scientific projects covering botanical, ecological, wood technological, archaeological, paleoecological and art historical disciplines. An increasingly important function of the collection is delivering scientific services for the monitoring of timber trade and commerce. Enforcing laws and regulations concerning forest management and timber trade requires taxonomic identification of timber. This relies on visualizing, describing and quantifying the rich information embedded in the wood anatomy of trees and comparing this information with reference material. However, timber identification based on wood anatomy is often a slow process requiring expert knowledge. In this respect, an increasingly important endeavour is the development of fast timber identification techniques to keep up with the pace of the growing needs of the international timber trade regulations. Over the last decades, the Wood Biology Service of the RMCA has explored a wide array of wood visualization techniques. This long-standing experience allowed to select a promising tool for fast timber identification: recognising macroscopic wood anatomical features through machine learning, using a large database of high-resolution optical scans of end-grain surfaces. A timber species identification tool based on the digitised images will significantly improve routine controls of timber shipments. Furthermore, by using machine learning on digital imagery, large quantities of wood anatomical information will become available in digital format. This data rescue process will valorise the collection and provide a much-needed basis for research in the context of biodiversity conservation and economic development.
Leading organization