3D-point-clouds-plant-phenotyping

Automated segmentation and annotation of 3D point clouds for plant phenotyping

File Structure

.
├── colored_annotation_processor.py   # Code for visualizing class and instance annotations                         
├── sample_output                     # Sample visualisation of class and instance annotation for ”A2_20220512_a.xyz“
└── README.md                         # This README file

Data

The dataset used is LAST-Straw, which can be found at LAST-Straw Dataset.

Running the file

Download the LAST-Straw dataset and place it under the ./data directory.

Make sure to install necessary libraries:

pip install numpy open3d colour-science

Then run

python colored_annotation_processor.py

Example

Example of an original .xyz file: ”A2_20220512_a.xyz“

Class Annotation

Instance Annotation

PointNet++ Visualisation

Avg IoU reaches over 70%