Graduate Project
 

Branch reconstruction and modelling for pruning

Public Deposited

Downloadable Content

Download PDF
https://ir.library.oregonstate.edu/concern/graduate_projects/mw22vf040

Descriptions

Attribute NameValues
Creator
Abstract
  • Successful pruning relies on accurately identifying the 3D structure of tree branches and leaders. However, this task is arduous in an agricultural setting due to the complexity of scenes, the presence of clutter, and variable weather conditions. This project addresses these challenges by leveraging advancements in 2D image segmentation and utilizing RGB and Depth data from a camera to generate precise 3D geometry for the tree. To achieve accurate tree structure identification, we employ a pre-trained MaskRCNN instance segmentation model trained on a smaller labeled dataset to obtain tree segmentation masks. Our algorithm leverages this mask and depth information to extract the geometry of branches and leaders, providing a comprehensive understanding of the tree’s structure. Subsequently, we use three methods to estimate the branch radius, enabling the construction of a detailed 3D mesh representation. This mesh is then reprojected onto the image, resulting in a refined segmentation mask that captures the intricate shape of thetree. The algorithm demonstrates remarkable performance, with a 97% accuracy in branch radius estimation and a success rate of 79.2% in generating masks for unseen data. These results showcase the algorithm’s effectiveness in handling the challenges posed by agricultural environments and significantly contribute to the advancement of accurate pruning in cherry tree farms and enhance overall crop management.
License
Resource Type
Date Issued
Degree Level
Degree Name
Degree Field
Degree Grantor
Commencement Year
Advisor
Committee Member
Rights Statement
Publisher
Peer Reviewed
Language

Relationships

Parents:

This work has no parents.

Items

Thumbnail Title Date Uploaded Visibility Actions
file details JosyulaGopalaKrishna2023.pdf 2023-06-15 Public Download