Computer Vision ID Label

[joellebel]

Here's some initial ideas. Are any of these resonating with anyone? I think I could live with B and F. I know originally I said we should stay away from the term "Computer Vision", but maybe we should just lay it on people. Open to other ideas.

Still working on the info text for the info blurb that is visible when the user hovers.

[loarie]

I think I like the robot icon first and the computer screen icon second. Not sure about the text

[tiwane]

I like the robot but it looks sad. Would it be possbible to take away the two vertical lines from its mouth so it looks more "neutral"? I like the blank computer screen second. The eye ones are freaky and I don't like the "magic" implication with the wand.

[joellebel]

@tiwane I can can turn that frown upside for ye.

Guys. I'm also wondering... what about folks who know what something is, but merely choose a taxon from the suggestions menu for convenience sake because that's the first thing in the autocomplete list. Wondering if we need to make some sort of distinction at the point of submitting the ID, kinda like how we now have this explicit disagreement thing. Any thoughts?

[loarie]

we can't distinguish that. this is an imperfect small change to imperfectly address a larger more complex issue

[naturalistcharlie]

i dunno if you want my opinion here since i have too many opinions but i think the robot is awesome and the eye is creepy. I do use the vision demo as a form of autocomplete sometimes but for me the option of knowing when something is from the algorithm is worth the downside of making it look like i am not coming up with some of my IDs. Given unlimited processing power (which we don't have of course) it would be neat to run the algorithm on all existing observations and have it stuck somewhere on the observation - not as a community ID factor and not even necessarily something you can hit 'agree' with (too easy) but boy would that be interesting. And it would be neat to see which observations were used to inform the community ID too. Ok i will stop now, thanks all :)

…

On Mon, Dec 4, 2017 at 3:37 PM, Scott Loarie ***@***.***> wrote: we can't distinguish that. this is an imperfect small change to imperfectly address a larger more complex issue — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#1560 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AG1tkxhIGlX9WcQF_Qwlrhx6mu_bTByDks5s9FgWgaJpZM4Q1Ejd> .

-- ============================ Charlie Hohn Montpelier, Vermont

[boucur]

pleary selected a Computer Vision ID [happy robot]
:)

[kueda]

Robot is a bit too Android-y / buggish to me, but I can live with it. Eyes are creepy. No eyes. I like laptop and wand.

[joellebel]

Here's some more options based on feedback. I kinda like the little little robot. I changed him a bit to match the style of our other icons which are solid, not outlined. (If I make him green he looks like an alien)

I'm also embracing the term "Computer Vision" more. I think if that's what it is, that's what we should call it and if folks want more details they can get it from the info blurb. "Computer Vision" soon will be a household name, no?

I think nailing down the text is the most important. So if we could do that, that would be wonderful.

Here's some spin offs on the above Direction C:

Here's one last idea which I like the best because it incorporates the ID icon with a touch of "magic" and we all know magic is just another way of describing science we can't explain. ...and we can't completely explain computer vision yet so it's quite magical.

[kueda]

I don't think anyone's objecting to the magic tab, and it seems fine to me. Is there an SVG available somewhere? "Computer Vision" for text is fine with me, but we'll need a tooltip to explain what that means b/c it's not obvious.

[joellebel]

Cool. I'll whip up the SVG now and base is on the same size as the ID label svg.

In terms of text, I took a crack at it, but definitely needs your wordsmithery. (I was trying to keep it as simple as possible.)

"Computer vision, as applied to iNaturalist, is the process of analyzing millions of identified observations to determine the key, visual traits of different organisms. These characteristics along with location data help inform a computer vision model or suggestion of a particular organism."

[kueda]

Ok, how's this: "Computer vision is the process of teaching computers to recognize patterns in images. We use computer vision systems trained on iNaturalist photos and observations to provide automated species identification suggestions. This label indicates when a person added an identification based on one of these suggestions."

[joellebel]

Luvit!

[joellebel]

@kueda the icon is ready and is here: https://www.dropbox.com/s/xvdvefnk0yx535j/icn_cv.svg?dl=0

I based the size off the ID icon so hopefully it will be harmonious with the font sizing.