Figure1. Refugee Camp mapping with using PulseSatellite
Counting and classifying structures in a refugee settlement is a common analysis task for humanitarian agencies. In practice, this is currently manually done by human expert analysts using satellite imagery. A single settlement may have tens of thousands of structures, and identifying each of them can take several days. For camp mapping, PulseSatellite uses a Mask R-CNN model (He et al. 2017) trained on images from 12 settlements, where the image of each has been split into 300x300 pixel tiles and annotated by human experts. Once the model has been run on an unseen camp, the analyst can inspect the result in the tool and correct the outputs on a subset of tiles. An adaptation stage can then be performed to fine-tune the model to the unseen image and increase performance. In recent independent tests camp completion rates increased from 77.3% to 94.7% after adaptation, with a final user accuracy of 94.4% -in line with humanitarian performance requirements described in (Quinn et al. 2018).
Figure2. PulseSatellite user-interface showing a hierarchy of learning models
* Note: The WG5 Best Practice "Mapping Refugee Settlements by using PulseSatellite" has extracted from a publication of Tomaz Logar et al. 2020.