christophersanmi-ux/Kaduna-school-mapper

AI project for Element of AI Final Task # Kaduna School Mapper Building AI course project ## Summary Kaduna School Mapper uses satellite images and a Convolutional Neural Network to automatically find and map schools across Kaduna State. This helps the government and NGOs plan internet access, funding, and emergency support for schools that are currently unmapped and underserved. ## Background In Kaduna State, especially in rural LGAs, there is no complete and up-to-date record of all schools. Without this data, some schools miss out on government programs, digital connectivity projects, and aid during floods or insecurity. * Problem 1: School location data is incomplete or outdated * Problem 2: Ground surveys are expensive, slow, and hard to do in rural areas * Problem 3: Without a map, it's impossible to fairly distribute resources like internet or books **Personal motivation**: I live in Kaduna and have seen how some rural schools are left out of digital initiatives simply because they are not "on the map". **Why important**: You can't improve what you can't measure. Mapping every school is the first step to ensuring every child in Kaduna has equal access to education and resources. Inspired by UNICEF + Development Seed's school mapping projects in Africa. ## How is it used? The solution is used by the Kaduna State Ministry of Education, NGOs, and telecom/internet providers. **Process**: 1. Download recent satellite images of Kaduna from Google Earth Engine 2. The CNN model scans the images and flags areas that look like school buildings 3. Human volunteers validate the flagged locations on the ground using a mobile app 4. The final verified map is published online and shared with decision makers **Users**: Government planners, NGOs, students and parents looking for nearby schools. **Context**: Used on desktop for planning. Field validation done on Android phones. *Example: Satellite imagery of Kaduna area used for mapping* ## Data sources and AI methods **Data Sources**: * Google Earth …