Penguin Watch Talk

Whether to classify things in the very far distance

  • l0b0 by l0b0

    On pictures like these there's clearly a mass of penguins in the background, but it's impossible to tell which tiny blobs might be rocks or other things. How sure do you want us to be about each individual blob being an actual bird before classifying? Where's the dividing line?

    In this case it would be very useful to compare the photo with other photos taken by the same camera; even better (but more work) would be to subtract another photo for example from the same time on a different day. Most likely any object which has moved (and therefore shows up as a noticeable difference) is a penguin.

    Posted

  • AvastMH by AvastMH moderator

    Hello I0b0. Thanks for your email. It's great to get some feedback.

    These shots are really tricky. We have some users with huge screens who can mark these with reasonable surety. But many who will be wondering rock or penguin? As you've detected, the terrain sort of works against us. It makes a 'dividing line' pretty much impossible. But the results we send in are acceptable so it's a case of 'carry on'.

    We had a long debate a few months back as to whether we should have more magnification (some people do have magnification programs on their computers which can help). But the scientists are happy with our counts, so we need to keep things steady and make no changes. I found that using more magnification just always left me with more '3 pixel' penguins just out of reach as it were.

    In certain pictures you will have part of the screen greyed-out so that our focus is on just one group. I wonder if later pictures of this site might end up with that change being made. We'll see once the scientists have got our earlier counts in.

    If you want to see the landscape-for-the-penguins what you need to do is keep a site in a 'collection'. When you look at the picture you see an identifying URL, just fiddle around with the last few digits to get other dates/times for the site.

    The important rule is that 'if in doubt please don't mark the object'.

    Thanks again for all your efforts - it's much appreciated I0b0.

    Joan (AvastMH)

    Posted

  • yshish by yshish moderator, translator

    Hey mods, I've just posted a 1st part of the FAQs under the Science board. Check it out please.. more coming soon! 😉

    Z.

    Posted