Kelp Talk

All Clouds

  • Rondine by Rondine

    What should I do when an image is all clouds? Should I mark the clouds or click next image so it can be removed from the dataset?

    Posted

  • Artman40 by Artman40

    Yes. The reason being is that the database knows that kelp cannot be seen even if there is some.

    Posted

  • Rondine by Rondine

    Thanks i'll just press next.

    Posted

  • Artman40 by Artman40

    Remember to mark "clouds" first.

    Posted

  • Rondine by Rondine

    Will do

    Posted

  • kkendle47 by kkendle47

    I just started classifying a few days ago. My interpretation of the instructions are that if I click NEXT IMAGE without marking a kelp bed, then the image is removed from the data set. So if the image only contains clouds with no kelp, is it still necessary to flag the image for clouds? I thought that the cloud flag was only important if I marked the image with kelp beds. Comments?

    Posted

  • yshish by yshish

    Hi guys,

    I want to ask what about the images which are misclassified as no kelp present.. Are they really immediately removed from the data set or need to be classified as NO KELP more than once to be removed (as at the other Zooniverse projects)? I actually have misclassified several of them since I had some technical troubles with my computer and was calming down myself repeating that 'There's no need to worry about that much, someone else will classify them correctly!'

    The same for the cloud button. It sometimes doesn't accept my clicking on the cloud button so the classification is recorded as NO CLOUDS.

    Thanks!

    Zuzi

    Posted

  • HMB6EQUJ5 by HMB6EQUJ5

    Hi good questions... welcome! I've wondered also about how often images need to be viewed before being retired on FF. As to the project it's a fun one and interesting images from LANDSTAT 4,5,7,and 8. You will find lots of "overland", "clouds" and image glitches. Hopefully a way to filter the land ones via lat/log will eliminate them prior. If you haven't try going through the blogs by the science team and also the discussion board ("Help"/FAQ) as they give more info. I'm clicking the Cloud button and then Next Image when I encounter and image with cloud cover. Sometimes I can see what i believe to be a kelp forest thru cloud breaks and then i just classify and don't press "Clouds". Best wishes... p.s. i believe i found some "Chats" with the science team on YouTube.

    Posted

  • Quia by Quia

    Looking at the blog, the number of people who have to say 'no kelp' is three, then the image gets put out of circulation. That's what it was when the project started, the number may have been tweaked since then, but I don't see anything discussing a change.

    http://blog.floatingforests.org/2014/08/08/the-image-policing-power-of-the-crowd-or-let-them-see-kelp/

    Posted

  • yshish by yshish in response to Quia's comment.

    Yes, that's what I read somewhere - that the count has been changed however I don't know when it should have be done and what the number should have been. Hopefully it wasn't after the blog was posted and if so, that the count isn't less than 2;)

    (Sorry for the grammar!)

    Z.

    Posted

  • DZM by DZM admin

    I'd need a scientist to confirm, but I can promise that the algorithms that we use always take into consideration that there may be disputing and erroneous classifications. 😃

    But we're also definitely trying to filter out bad/cloud images as quickly as possible!

    Posted

  • jebyrnes by jebyrnes scientist

    We have a lower threshold for no kelp images than kelp images. One 'next' with no kelp isn't enough. Right now the threshold for removing an image is three 'no kelp' classifications. See http://blog.floatingforests.org/2014/08/08/the-image-policing-power-of-the-crowd-or-let-them-see-kelp/ for more.

    Posted

  • yshish by yshish in response to jebyrnes's comment.

    Thanks for your answer, good to know:]

    Z.

    Posted