Accuracy assessment of game-based crowdsourced land-use/land cover image classification

Crowd-sourced data has been used as alternative for ground truth data. However, the accuracy of crowd-sourced land use and land cover classification information has not been explored in an African land cover and land use context. The compares traditional classification techniques to crowd-sourced classification, highlighting the high accuracy of crowd-sourced responses and short turnaround time compared to other supervised and unsupervised methods. Crowd-sourcing is found to be an acceptable method for high speed classification of land use and land cover, and further study is recommended.