Over the past two years, HOT has taken a leadership role in exploring and facilitating the application of the latest advances in artificial intelligence.
The main aim has been to find best ways to amplify the impact of humanitarian mapping and assisting our growing community of mappers all around the world that contribute their valuable time to build the largest open geospatial platform.
Collaborating with the OpenStreetMap community and the best machine learning engineers from Development Seed, Facebook and Microsoft, we are piloting several artificial intelligence applications in the field of machine learning for humanitarian mapping. These pilot applications and related mapping results allow us to now evaluate the benefits and potential limitations of the assisted mapping approach in comparison to traditional mapping.
We are conducting research to quantify the measurable impact of an AI-assisted mapping workflow and we believe a reproducible and transparent study will give us a clue.
What we want to compare:
Well-mapped OpenStreetMap data (over a longer time period)
Traditional OpenStreetMap mapping data (single remote mapping event)
AI-assisted mapping data with RapID (single remote mapping event)
The data created by the latest generation AI models (without any editing)
What we want to measure:
Quality: Existence of the objects and similarity of the polygons
The speed of mapping
We want all the data and workflows we produce to be as reusable as possible. Please reach out to us if you are interested in any further analysis, we are happy to hear your suggestions before we start, so we can assure all raw data is as useful and correct as possible.