This prototype requires Python 3 with the skimage
package (and matplotlib
)
installed. Type the following to launch the demo:
git clone https://github.com/PillarsOfCreationTeam/scoring.git
cd scoring
python3 -m http.server --cgi 8000
Then, the demo will be available at http://localhost:8000
.
This is a scoring system prototype for the Galaxy Quest game (see main repo). In this game, users will help NASA to identify galaxy boundaries and nebula axes, i.e., tasks that may be difficult for AIs, are quite human readable.
The scoring system is based on the wisdom of the crowd. For each Hubble image, the backend stores a mask that is a moving average of the individual draws. The game rewards users that are able to match their draws with the masks, thus earning Hubble time to continue the game exploration.
For example, this is the status of the mask for the NGC 5972 galaxy after a test with 52 participants:
In a real implementation, there could be a tier-based, or reputation-based, weighting system in which best performing players would have more influence on the mask of new images.