Wood is a plentiful and renewable resource extensively used in construction and manufacturing industry, including furniture production, musical instrument crafting, and paper manufacturing. As a natural material, its mechanical properties vary due to the distinctive anatomical features inherent in different wood species. The exploration of the relationship between anatomical features and the mechanical behavior of wood remains an ongoing subject of research.
In this presentation, two recent studies will be introduced to showcase the potential of artificial intelligence (AI) approaches in elucidating this relationship. The first study focuses on cell deformation analysis using computer vision techniques. A deep learning-based semantic segmentation approach is successfully applied to partition individual wood cells in cross-sectional image. Additionally, by combining a particle tracking algorithm, it becomes possible to evaluate the deformation of thousands of cells simultaneously during mechanical testing (Figure 1).
The second study applies machine learning approaches to predict the mechanical properties of wood. The convolutional neural network was used to accurately predict the modulus of elasticity (MOE) and modulus of rupture (MOR) of wood in the transverse direction based on its cross-sectional image. Moreover, AI explainable technology was used to clarify important features located in the cross-sectional images that are highly related to the predicted result (Figure 2).