Purdue University Graduate School
Browse
- No file added yet -

Selected Topics of Deep Learning Application in Forest Research

Download (14.77 MB)
thesis
posted on 2021-12-19, 19:44 authored by Fanyou WuFanyou Wu
Digital Forestry uses digital technology and multidisciplinary expertise to measure, monitor, and manage urban and rural forests to maximize social, economic, and ecological benefits.

In chapter 2, we investigated the potential use of CNNs for hardwood lumber identification based on tangential plane images. In chapter 3, we developed deep bark, a lightweight tree species identification application, by using deep learning. In chapter 4, we first introduced a new dataset of images of hardwood species annotated for tree ring detection. We applied the state-of-art semantic segmentation models to the dataset. In chapter 5, we combined the observed classes and non-observed classes by distinguishing the attributes of objects and applied zero-shot learning to microscopic wood images.
The results above chapters demonstrated the potential and effectiveness of machine learning in many forestry-related tasks. Those applications help both the research community and industry to conduct better digital forestry business. However, we still need to point out that the availability, quality, and quantity of data and annotation are critical factors in conducting meaningful research and applications in forestry.

History

Degree Type

  • Doctor of Philosophy

Department

  • Forestry and Natural Resources

Campus location

  • West Lafayette

Advisor/Supervisor/Committee Chair

Rado Gazo

Advisor/Supervisor/Committee co-chair

Eva Eaviarova

Additional Committee Member 2

Bedrich Benes

Additional Committee Member 3

Henry J. Quesada

Usage metrics

    Licence

    Exports

    RefWorks
    BibTeX
    Ref. manager
    Endnote
    DataCite
    NLM
    DC