Kelp Talk

Recommendation ideas for classifying

  • Hemipelagic by Hemipelagic

    i have a few concerns with the current classifying interface. One, there are instances when artifacts or errors (like those yellow/red/black bars) are present, yet so are kelp forests in the picture. I'm not too sure if I should skip it completely or still mark it.

    Also, the "cloud" option is a little vague. Should we only click it when there's a chance that clouds could be potentially covering kelp forests? So if there's an image of a coastline with any and all clouds completely over land and no clouds over the ocean, is it still worth marking as clouds?

    At any rate, I have a recommendation for the interface.

    1. As the satellite image shows up, users are given the option to choose "coastline", "land", "open ocean", "full cloud cover" or "other/cant tell"

    2. After clicking any of the options, the user is asked if there are any artifacts that are covering up part of the image

    3. If the user first clicked "full cloud cover" or "land", then they are taken to the next image (since no additional/valuable data can be extracted from that image)

    4. If the user clicked "coastline" or "open ocean", they are asked to mark any kelp forests present (or they can click the "none" button)

    5. After marking, the user is asked if there are any clouds that could potentially be covering kelp forests

    Any thoughts? I hope this doesn't come off as nit-picky or anything. I think this project is fascinating and I just want to make sure its as successful as possible!

    Posted

  • DZM by DZM admin

    Thanks for the feedback, @stanley.vissil ! I'm not sure if the project is up for any sort of redesign before the next dataset arrives (whenever that will be), but if so, I'll be sure to point to your suggestions. Glad you're enjoying the project!

    Posted