Weddell Seals (Last?) - Strange colored polygons


#21

This one goes with the one @Shaul identified on 108964

(http://www.tomnod.com/campaign/campaign_2284/question/108965)


#22

(http://www.tomnod.com/campaign/campaign_2284/question/105419)


#23

I am surprised how many are showing up…
(https://www.tomnod.com/campaign/campaign_2284/question/44965)


#24

another one…

(https://www.tomnod.com/campaign/campaign_2284/question/45306)


#25

Yes, there are a lot strewn around the Antarctica! Maybe @Mel_Nod will come up with something more, to solve the mystery.


#26

You sure seem to be seeing a lot of those strange coloured blocks, Terri! I haven’t seen a single one yet in the tiles I’ve done so far… only the ones posted here!


#27

I agree. I am surprised how many are still showing in this final campaign, but as Mel pointed out they are built into the image. Next week when she is back we may hear more about them. :smiley:


#28

I think I remember Mel saying something about trying a different method of image processing, which has slightly “overdone” it and has left the strange coloured artefacts scattered all over that part of Antarctica…

LOL, perhaps I should feel slightly “left out” because I haven’t been presented with any tiles with the strange colours in my searches through the image queue… :thinking::upside_down_face::wink:


#29

http://www.tomnod.com/campaign/campaign_2284/question/45310


#30

@Mel_Nod 225413 also shows this yellow polygon


#31

Thanks all. Nothing I’m about to say here “solves” the issue - that would mean me re-processing all the images and starting over the campaign, and we already have 76,000 votes :tada::sparkles::raised_hands::full_moon_with_face:

But it’s a little interesting, at least to me.
Here’s a piece of an image from Antarctica that has been hand-processed by our team in Longmont:

46%20PM

And here’s some detail from that same image after it has been given the basic, automated processing steps that are given to most images (not the by-hand treatment):

The colors are way off, and generally in those shades of blue, pink and yellow that we are seeing in the campaign. We aren’t totally sure what has gone wrong, but it probably has to do with how far south we are, the angle of the specific satellite collection, the features of the area that’s being collected, and the general difficulty with our imagery processing so much white. My guess is that the images we are seeing the colored polygons on have this discoloration and the others don’t (I am going to do a little more testing).

We don’t have the bandwidth to hand-process our images for this project but have developed many automated ways of getting good, somewhat realistic color into our Antarctica imagery. We tested some new methods this last round, and I think some relics of this original discoloration are left in our image (but I do think the contrast/detail is pretty good where we are not seeing these polygons).

Not positive this is the problem, but it was very illuminating for us to see this image partly-processed before it went through the rest of our workflow.

Long story short - it’s been difficult making all aspects ideal in Antarctica, between colors and details and light and shadow!


#32

oooooOOOOOoooooo I see a seal !

image
.
@EmeraldEyes Do you see him? her?

(scampering away before Mel swats me with a DigitalGlobe blog post!)


#33

We don’t know how you do it at all !

I would like to know what causes the look of rectangle depressions on Antarctica? Or giant depressed foot-prints? Coolest… but what makes them?


#34

image running%20away

:rofl:


#35

I’ve been finding quite a few too, some probably already posted on here, but as @Mel_Nod has a fair idea of the “why” now and the fact that they don’t actually interfere with the task in hand (with only one actually inside a poly so far), I haven’t bothered posting them anymore.


#36

Thanks for working around these, folks. Never a dull moment with our campaigns.

:laughing::joy: @cageycat @EmeraldEyes


#37

image

I think I saw that bloke before…

Or maybe it was that other bloke. lol


#38

I showed my wiz-kid GIS Analyst your pretty colored picture and she starts explaining image processing to me. And how many color channels different satellites use… Only thing I knew is that infrared reds are healthy trees and vegetation.

Wish she’d apply at DigitalGlobe. :smiley:

By the way, I was wondering…Could DG write a program that says, for example,
All Lime Green is White
All Blue Green is Gray-ish
All Orangish is Brown
All Pink is… Party-time!


#39

:laughing::laughing::laughing::rofl:


#40

Haven’t seen these for a while…

(https://www.tomnod.com/campaign/campaign_2284/question/99949)