How It Works

A Picture

Geosurvey crowdsources the collection of scientific data about our planet. It presents you with a random image of the Earth, which may be taken from satellites, drones, or mobile phones. Your task is to answer questions about that square, such as assessing its cultivation level, cropland boundaries, woody coverage, or prevalence of plant disease.

The questions you see are specified by the survey designers, which are scientists from universities and research organizations that we have partnered with. Their questions can be arbitrarily customized, including conditional logic and online instructions. Likewise, the regions of interest for each study are chosen by the designers through shapefiles and related geospatial specifications.

Pan Controls, Zooming, and Layers

To get a better view of the area of the Earth being displayed, you can use pan and zoom controls. You can also pull images from different satellites (changing the base layer) to look at the region through a variety of lenses.

Your Move

On the right hand side of the screen, you can choose to answer questions about the region inside the square, or skip and be presented with a brand new square. Also, if you ever lose track of the square while doing your panning and zooming, you can click on "Refocus map" to recover the original camera orientation.

Selecting the Best Option

The most common kind of question asked by survey designers is a multiple choice question. After playing with the aforementioned controls, select the choices that best describe the scene strictly inside the box, and then click "Submit." Or, if you are unsure, you are encouraged to click "I don't know." You can also use the ``Discuss" panel to talk about what you see with others.

Submit and Climb the Leaderboard

Your submissions are tracked and credited toward your score. The more questions you answer, the higher you will climb in the leaderboard, and the greater your chances will be at winning a prize! Your helpful answers will be used to assist agricultural workers and governments with producing more accurate land usage maps, and for training computer vision algorithms to produce more accurate inferences of satellite imagery content.