Over the past few months HOT has been working in partnership with NESTA Collective Intelligence grants and 510 at the Netherlands Red Cross to implement an experiment to test the efficiency and efficacy of AI assistance on remote mapping of buildings.

The experiment is designed to compare the results of traditional remote mapping workflows (editing in ID Editor) with emerging AI assisted workflows (editing with RapID). To do this, we will be conducting mapping experiments of two locations (Uganda and US), with two different levels of mappers (beginner and advanced) using the two different remote mapping workflows (RapID and ID).

To this end, we are looking to run four experimental groups with > 50 people each, as follows: Beginner Mapathon - Uganda data Beginner Mapathon - US data Advanced Mapathon - Uganda data Advanced Mapathon - US data

Each mapathon will be a total of two hours, with participants being randomly assigned to map building footprints with either ID or RapID prior to beginning their mapping. Data will be gathered on the existence of map features, completeness of mapped features and similarity of map features, when compared with an OSM reference dataset.

For the mapathons we will be using convenience sampling from our networks by generating a public call for participants.

You can sign up to participate in the experiment here >

Kindness,

bo