Map land cover use and building footprints at national scales.
Deliver scalable raster and vector maps online.
QED uses artificial intelligence and satellite imagery to enable governments, international development organizations, and the private sector to make sense of land-use at regional and national levels.
Our most commonly requested land use classifications are national-scale maps of croplands and building footprints, which are crucial for applications such as agricultural intensification, vaccine delivery, and disaster relief. These maps allow front-line workers to optimize logistics, and enable policy makers and entrepreneurs to better quantify national needs and allocate resources accordingly.
Throughout the Global South, governments and businesses have struggled to generate these land use maps. And running in-person surveys with large ground teams is extremely time-consuming and expensive.
By coupling deep learning with innovative methodologies for the crowdsourced collection of training data, QED has been able to produce land use maps at national scale while cutting the time and cost of generation by many orders of magnitude.
Our maps have been generated for both public and private sector clientele, including The Nature Conservancy, the USAID Feed the Future Program, Visualizing Insights for Fertilizer in African Agriculture, and the Africa Soil Information Service.
We present maps over the web in a scalable-fashion, supporting both raster and vector formats at multiple zoom levels. Following the slogan that “Seeing is Believing”, our maps can be inspected for their veracity, resulting in information that is often more up-to-date or reliable than government statistics.
Nigeria has the largest population and economy in Sub-Saharan Africa (SSA). With the recent plunge in the price of oil, Nigeria is urgently seeking strategies to diversify its economy. Strong attention is being placed toward Nigeria’s agricultural sector. Better availability of reliable and up-to-date agricultural data will help both the government and private industry to inform policy and investment in the agricultural sector. And basic statistics on the hectarage of cropland and land area by major crop type are outdated, if they exist at all. Below is an example of cropland mapping, generated by QED across Nigeria, to fill this gap. Please use the interactive slider to compare satellite imagery against our classifications.
Plot boundary mapping are another area of active research that we are pursuing and have promising initial results, but which require larger datasets to effectively extrapolate our automated segmentation models across the country. Below are some examples of our work in this area.
Effective management of farmlands, grazing reserves, forests, irrigation, wildlife habitats, and endangered species requires the classification of land cover at national scales. Using machine learning, we are able to scalably generate maps such as the one below, differentiating between croplands, urbanization, forests, shrublands, and rivers and estuaries. Please use the interactive slider to compare the satellite imagery with our classifications.
Many countries in Sub-Saharan Africa can benefit from greater access to electricity. Mapping of building footprints at national-scale, particularly in rural areas, can assist companies with planning electrical layout and utility placement. Below are some examples from Malawi, which we have mapped in its entirety.
Updated soil information is extremely scarce throughout the world, despite the fact that soil is one of the most precious natural resources of any nation. This information is acquired at great economic and temporal expense, requiring both field work and analytical chemistry. Through the AfSIS project (2009-2019), we have expertise in pairing soil data with machine learning and remote sensing to make digital soil maps.