pywood21 / prop_pred_2023 Goto Github PK

The convolutional neural network (CNN) was applied in this study to automatically extract the anatomical features of wood for predicting their mechanical properties in the transverse direction.

And this repository aims to show the workflow for training regular CNN and density-informed CNN for predicting the modulus of elasticity of wood with different orientations of ray parenchyma cell (ORP).

You can find more information in:

Shuoye Chen, Rei Shiina, Kazushi Nakai, Tatsuya Awano, Arata Yoshinaga, Junji Sugiyama,. "Potential of machine learning approaches for predicting mechanical properties of spruce wood in the transverse direction", Journal of Wood Science, 69:22, DOI：10.1186/s10086-023-02096-z　(https://rdcu.be/dfl55).

The repository contains the necessary codes for building two CNN models to predict the mechanical properties of spruce wood in the transverse direction from their cross-sectional images. For the details please check the jupyter notebook:

CNN_model_building.ipynb