plot-boundary-extractor module

This module is to extract field experiment plot boundaries for agricultural applications. This module was initially developed for wheat breeding trials, but it can also extract plot boundaries for other row crops. More tests will be done and shared in the future.

This repository is associated with our research on Developing a segment anything model-based framework for automated plot extraction, which has been submitted to Precision Agriculture journal and is currently under review.

Citation

If you use this repository in your research, please cite the following paper:

Kim, H.S., Olaniyi, I., Chang, A., & Jung, J. (2025).
Developing a segment anything model-based framework for automated plot extraction.
Precision Agriculture, 26(3), 53. Springer.

@article{kim2025developing,
  title={Developing a segment anything model-based framework for automated plot extraction},
  author={Kim, Han Sae and Olaniyi, Ismail and Chang, Anjin and Jung, Jinha},
  journal={Precision Agriculture},
  volume={26},
  number={3},
  pages={53},
  year={2025},
  publisher={Springer}
}

This module was developed by Geospatial Data Science Lab at Purdue University. If you have any questions, you can create an issue in this repository or contact Hansae Kim (kim4012@purdue.edu) or Jinha Jung (jinha@purdue.edu).

plot-boundary-extractor installation

Set up a virtual environment

First, we need to install plot-boundary-extractor on the machine you want to run this example tutorial. We will set up a new Python Virtual Environment called pbe.

$ conda create -n pbe -c conda-forge gdal python=3.10 -y

We need to activate the newly created virtual environment before we start.

$ conda activate pbe

Install plot-boundary-extractor module

Now we need to install the plot-boundary-extractor module using pip.

$ pip install plot-boundary-extractor

Install torch

The plot-boundary-extractor module uses torch module, so we need to install it on the machine you would like to run this example tutorial.

CPU version

If your machine is not equipped with NVIDIA GPU, then you will have to install the torch without GPU support.

$ pip install torch torchvision

GPU version

If your machine is equipped with NVIDIA GPU, then you will have to install the torch with GPU support.

$ pip install torch torchvision --index-url https://download.pytorch.org/whl/cu118

Install SAM

The plot-boundary-extractor uses SAM to detect plot boundaries automatically, so you must install SAM using the below command.

$ pip install git+https://github.com/facebookresearch/segment-anything.git

You also have to download a pre-trained model using the below command. Please make a note of where you're downloading the pre-trained model. We will need this path information in the later tutorial.

$ wget https://dl.fbaipublicfiles.com/segment_anything/sam_vit_h_4b8939.pth

Now the installation is done, and you're ready to extract plot boundaries using the plot-boundary-extractor module. An example Jupyter Notebook is available for your reference.