Company-Project:
IIASA - Picture Pile/Crowd2Train
Description:
We will present the picture pile classification platform which allows to collect reference data for remote sensing application as well as generic classification tasks for which annotated or labelled data are needed.

Based on either very high resolution satellite data collected from Google Earth or Bing Maps via their APIs or through ground based or Street level photography a rapid and effective classification of those images can be undertaken. The reference data or annotated images can be used by ML algorithms to perform this task automatically or as input data for remotely sensed classifications based on free and open data such as Sentinel 2. The classification can either be binary, continuous or categorical.

5 different applications will be shown which use different input imagery namely: crop type labelling based on Streetview/Mapillary (crowd2train project), deforestation monitoring, poverty mapping (Game.EO project), ocean plastic monitoring as well as early disaster response. In particular applications which demo its use for SDG monitoring will be shown.

The platform will allow anybody to submit tasks and the crowd can also be recruited through a payment scheme which has been added to the Picture Pile Platform (PPP).