Dwelling Detection on VHR Satellite Imagery of Refugee Camps Using a Faster R-CNN
The management of humanitarian operations in highly intense situations like migration movements happening at borders often lack current and suflicient information. Satellites do provide large-scale information fast. When dealing with a migration situation, satellite images now can give information about where refugees are before they arrive at a border, giving first responders urgently needed lead time for contingency and capacity planning. Dwelling Detection, a method conducted on satellite images of refugee camps, is able to count the dwellings in a camp. From that, the number of inhabitants in a camp can be estimated for forecasting purposes. To count the dwellings, object detection machine learning methods can be used. Inthis paper, a dwelling detection workflow using a Faster R-CNN i s described To train the Faster R-CNN, a fast training data annotation workflow was developed. The Faster R-CNN outputs an estimate of people living in a camp and a con.fidence factor, giving a global evaluation metric about the quality of the analysis of the image andby that of the calculation itself. This workflow yields results that can be used in humanitarian operations. So our related proposal is to get satellite images fast, evaluate them with our method, and have better numbers für contingency and capacity planning. By this, stress für all people involved in a humanitarian (crisis) situation can be reduced.