Speaker
Description
The identification of standing tree species, forest timber assortments, and on-site as-sessment of their quality and value using images would be widely useful for a variety of applications. In this study, we used the Orange data mining software to classify images of forest wood assortments by tree species and value. The software applies deep learning algorithms for image classification. Our results show high accuracy when classifying assortments by tree species, but lower accuracy when classifying assortments by market value. Classification accuracy for spruce assortment images by market value was only 39.1%. We conclude that further exploration of approaches to classify assortments by value is needed, as this task has the greatest practical significance but currently yields the least reliable results.
| Keywords | classification; quality; machine learning |
|---|
