This project is solving the My Sky Color challenge. Description
Skynet is a crowdsourcing platform whose operations are almost exclusively performed by smartphone-toting humans, in service of its goal of furthering our understanding of air quality, aerosol loading, and our measurement of said.
Sky color is an indicator of air quality; specifically, it provides a qualitative indication of aerosol loading in the atmosphere.
A bright and clear blue sky means the air is clean; a white and milky sky means the air is not so clean.
The intent of this challenge is to combine the qualitative "the sky looks blue today" with the quantitative "how blue is it exactly." We allow the user to select which known shade of blue the sky most closely matches, both so that they feel like they are "doing science" and so we have a cleaner dataset. We also record the time and location of the photo, for future analysis.
We are not currently comparing these results to Aeronet's measurements, but that is a clear direction for the future.
The Skynet app is responsible for taking photos, analyzing pixel data to determine the dominant color palette in the photo, selecting the colors which are closest to our known shades of blue, and prompting the user to select the best match and then submit his/her selection.
It is a mobile web app, written in HTML5, which is compatible with iOS and Android. (It can be installed to the home screen of both platforms, and act as if it is a real native app. Except slower, especially for the CPU-intensive photo analysis.)
Currently, the platform itself is extremely basic.
It is a Ruby on Rails app, whose database has a single table to track submissions (the photo, the selected blue factor, the location, and the time of submission).
It serves primarily to support the app and to record the global submissions.
- Display photos that have been submitted near the user's current location
- Display the aggregate air quality over time for a location radius
- Provide users with the ability to "vote" for good submissions (i.e. users vote for a given photo when it appears to actually be a photo of the sky and the selected blue factor appears accurate)
- Send users' top-voted photo submissions to Aeronet or other related networks. Our logic here being that by socially validating submissions through user votes, we are more likely to submit quality data to Aeronet.
License: Apache License 2.0 (Apache-2.0)
Source Code/Project URL: https://github.com/sirsean/skynet
Mobile Web App - http://skynet.photos/