Pseudo Absence Generation and Locust Prediction

Predicting locust breeding ground locations from satellite data.

Research Paper

Install

Create a virtual environment and install requirements.

pip install -r requirements.txt

Notebooks and scripts

Run the notebooks with Google Colab or appropriate Docker container.

Notebooks	Link
Colab Intro (Python)
Pseudo-Absence Generation (R)	View
Pseudo-Absence Generation Viz (R)	View
Presence-Only (MaxEnt) Data Generation (R)	View
Presence-Only (MaxEnt) Modelling (R)	View
Add Environmental and Climate Data (Python)	View
Model Training (Python)	View
Model Interpretation (Python)	View
Hypothesis Testing (R)	View

Docker

Build the image running the following.

make build

Start a docker container in bash

make bash

To launch a notebook use make run_notebook.

For the R Docker Container add version=r to the build and run commands.

Download Preprocessed Data

Download and extract the preprocessed data into data/ directory from here

Preprocess Data

To run the preprocessing workflow, the following datasets are required:

• FAO Locust Observation Data
• NASA GLDAS_NOAH025_3H Dataset
• ISRIC SoilGrids (Refer to this notebook on how to download SoilGrids data)

Run the following notebooks sequentially, to generate preprocessed data

1. Pseudo-Absence Generation. You can run Pseudo-Absence Generation Viz for visualization.
2. Add Environmental and Climate Data

Citing

If you find this project useful in your research please consider adding the following citation:

@proceedings{yusef2021locust,
    title     = {On pseudo-absence generation and machine learning for locust breeding ground prediction in Africa},
    author    = {Ibrahim Salihu Yusuf and
                 Kale-ab Tessera and
                 Thomas Tumiel and
                 Sella Nevo and
                 Arnu Pretorius},
    journal   = {Advances in Neural Information Processing Systems (NeurIPS) workshop, 2021, Sydney},
    year      = {2021},
    url       = {https://arxiv.org/pdf/2111.03904.pdf},
}

NeurIPS 2021 Workshops:

• Artificial Intelligence for Humanitarian Assistance and Disaster Response (AI+HADR)
• Machine Learning for the Developing World (ML4D)