EVE General Discussion

Yea, I always feel like a chump clicking cell-to-cell variations, even when it is obviously the case. I just do it anyway.

With that one, I'd have chosen the 92%, and there's cellular wall stuff going on because in the center left-ish area if you turn off the red and the blue you can still see the shapes of the cells.

I think I figured out the 92%: you switch to just green and blue, and if you can still see the potato with holes shape with the blue turned off and just green showing, then that's the 92% choice, otherwise if you just see the shape of the nucleus but not the darker holes, then it's the 7% choice.

With "unspecific" you can't tell, but in the case of "cell to cell variations" I believe that you can't just pick that, you have to also pick the pattern that varies, or the 2-3 different patterns that you see. It's like:

"Which of the following patterns do you see here?"
"Yes."

There can be multiple classifications for one cell, so you can have, for example, selections from the nucleus and cytoplasm sections just like you had in the tutorials. Anything that should be mutually exclusive excludes you from making any conflicting choices.

With that image, the staining is more concentrated in the nucleus and you can see "holes" in it that line up with the "holes" in the blue so the 92% are correct.

The staining in the cytoplasm (red bit) could be background staining or could be something else - hard to say without being able to change colour channels.

Personally, I would be considering selecting cell-to-cell variation as well due to the different intensities present.

In any case, it is not unidentifiable because there is clear differentiation in staining across the cell and there are clearly identifiable features.

Sample #100054449

Blue - Green - G+B

I feel like a rebel... or a pioneer.
Cytoplasm (80%), sure... but can everyone miss small spots "overlapping with holes in the blue marker"?

I had trouble tagging with "nucleus" (80%) because staining intensity seems at the same level both in and outside the nucleus.
I regret not tagging "nuclear membrane" - its kind of faint, didn't notice until now.

That's the Jove nebula 0.216 secs after supernova from a distance of 15.764 LYs with optical enhancements.

Hi Dichroic,

Thanks for the feedback, interesting stuff -). Yeah with the recognition filters what i was thinking was areas where for e.g. you have a straight line the software measuring that. Or if there are circles that can be measured. Measuring patterns or pattern recognition. This also calculates the light or colors, so for e.g. we have a circle that is recognised and we also measure the light. Thats why i asked before do specific patters have a set ratio based on color light scheme they fall in.

I'm probably going way off track with the game here. As you know military jets scan on heat, so when they fire a rocket it's going towards the heated area. Thats just an e.g. I dont want to come up with something that makes the game boring. Just trying to help. Yes it will eventually come down to the user filtering it out. The computers can only give us the scanned patterns or the most matched and then we decide.

So my dream feature in this game probably asking too much was going to be, can we have a filter that the cpu does most of the calculations to around say 80% and then the remaining 20% be given to the human to make judgement on. What we are given so far is color tools. We dont have tools to use to measure or work with patterns even those color tools do show some patterns in a way. Yet some basic photos i've worked with have been wrong because of patterns and not colors. From what i see you think you get it spot on because the colors are accurate with the supported e.g. photo in game.

So far we have talked about patterns and light. Have we talked about sound vibrations -) does sound affect these cells differently when they are affected by sound -). That's one form of pattern recognition only if they all react differently to it for e.g. circles react differently to straight lines when sound or vibration is applied. Anyways too much chit chat from me.

The best references for light measurements and temperatures would military oh and let's not forget nasa. Nasa as you know use specific measureing tools to measure pluto's surface. Also xray the surface. Just a thought. Anways thank you for the info. Will see how things progress so far it's fun and new.

Many thanks.

I started off really bad. Even though I was trying, button mashing would have probably gotten better results!!

But I clawed out of a 40% accuracy at level 12 and back at 50% at level 20. Feels good. I want that armor lol

I need to get an image hosting thing, I've had some really interesting ones. One looked like a nucleus popped and all the goo came out! It was glowing bright green and there was a black spot where the nucleus should have been.

Would love to see that one! :)

I'm a bit confused about this one. I might be totally wrong but when I read the description of cytoplasm it says:

Seen throughout all the whole cell, except in the nucleus (blue marker). The intensity can vary throughout the cell, and is often stronger close to the nucleus.

in other words: all the red might be green but the blue is blue without a spec of green in it.

Now I got image 100096917 (as in the screenshot here: http://i.imgur.com/cVv5kwc.png ). Basically its green all over so I figured the sample was useless. But to my surprise 50% match was on cytoplasm. Am I misunderstanding cytoplasm and reject my samples to soon. Or did 50% of the people not read the first line of the description?

EDIT: same on 100021233: http://i.imgur.com/wRnL8Sq.png that one is even fully conflicting with both Nucleus and Cytoplasm