Identifying School Locations in Sudan from Satellite Imagery
Giga is a joint global initiative by UNICEF and ITU to connect every school to the internet and every young person to information, opportunity and choice.
Giga has teamed up with the world’s first and biggest open-source AI4GOOD library, OmdenaLore, to develop an AI model to identify school locations in Sudan using Satellite Imagery.
The problem
Accurate data about school locations is critical to Giga, a joint initiative by UNICEF and ITU aimed at connecting the unconnected schools in the world to the internet. This will help bridge the digital divide in the world. However, for many countries, school location records are often inaccurate, incomplete, or non-existent. Traditional methods of the field visit and mapping of the school locations are not only heavily expensive but some schools are located in remote, inaccessible, and insecurity-prone areas.
Therefore, the mission of this AI project is to develop a Deep Learning model(s) to accurately and comprehensively identify school locations in Sudan from high-resolution satellite imagery.Â
The project outcomes
Omdena ran the project as a six-week challenge involving 52 collaborators. Participants were given a geo-diverse, high-resolution satellite image of Sudan and some school location data points. The team trained the AI on how to distinguish between schools and other types of buildings. The resulting model was able to determine the precise location of schools and their boundaries with 95% accuracy – a high rate for this type of model.
Many children in Sudan are not able to enjoy the benefits of online learning as school connectivity levels are very low. Mapping can help address this problem by providing insights for the planning and roll-out of connectivity infrastructure. The Omdena exercise will therefore contribute to improving the state of education in Sudan which is affected by challenges like the high number of out-of-school children, a lack of teaching materials, and inadequate facilities.
The UNICEF Sudan Country Office played an important role in this exercise. With a presence on the ground and a deep understanding of the context, the Country Office was able to provide the AI mapping team with crucial parameters to help identify schools from the satellite imagery.
The next phase of the project is to apply the new AI model to the entire country, thereby uncovering more schools with unknown locations. Mapping will support the Government in knowing precisely where resources are needed and assessing the reliability of internet connection where it already exists. These are vital steps towards the goal of giving every child access to online learning opportunities.
Read about OmdenaLore
We went on a mission to build OmdenaLore, an open-sourced data science package that provides comprehensive and ready-to-use Python classes and functions to solve almost any machine learning problem in an accelerated manner. This Python library is built and maintained collaboratively by the global AI community, thus making it more inclusively and ethically developed.
Requirements
Good English
A very good grasp in Computer vision and deep learning concepts
Student, (aspiring) data scientist, (senior) ML engineer, data engineer, or domain expert (no need for AI expertise)
Programming experience with Python
A strong grasp of Git concepts and Git workflow
Application Form
Become an Omdena Collaborator
