It is our pleasure to announce that our project “Namibia, a new dataset for rangeland and pasture management” has been selected for funding by Lacuna Fund. Lacuna Fund is an American Institution that supports the creation, expansion, and maintenance of equitably labeled datasets that enable the robust application of machine learning (ML) tools of high social value in low and middle income contexts globally. The call, named “Labeled Agricultural Datasets for Machine Learning Solutions in Sub-Saharan Africa” was launched in 2021 and aims at “providing funds to create, expand, and/or maintain datasets that fill gaps and reduce bias in labeled data used for the training and/or evaluation of machine learning models as well as to fund the development and maintenance of equitably labeled datasets most effectively and efficiently.”
Namibia’s rangelands and ecosystems are under serious threat, their erosion and level of degradation are posing a risk on local biodiversity and food security. Overgrazing, frequent and prolonged droughts as well as lack of adequate land management resources are among the culprits. Moreover, climate change and bush encroachment decisively worsen the problem.
This pioneer project developed by Farm4Trade Namibia, with the support of the University of Namibia (UNAM) and other local partners, aims to create a new dataset that can lead and support the development of forecasting Machine Learning models to improve pasture management strategies and simplify decisions on the carrying capacity of selected areas.
The satellite imagery will be paired with their respective approximate land-cover labels. Remote sensing imagery will be annotated in collaboration with the local experts who will conduct the field activities on a monthly basis to observe the type of vegetation present in the selected areas. The collected data will include the vegetation stage, quantity estimates of perennial and seasonal plants, vegetation type, dominant plant species and the degree of grazing or browsing pressure. Data on livestock density in the study area will also be collected and associated with the labeling of satellite imagery.
The project aims to bring benefits to multiple stakeholders of the agricultural and livestock sector and in particular to:
The creation of the aforementioned dataset will also allow Farm4Trade to apply its AI technologies to develop an innovative pasture management application able to automatically forecast grazing and pasture quantity and estimate the carrying capacity of a selected area. Overall, the data will empower the community to better manage, plan and mitigate rangeland health.