pjeffers51 Loc: Richmond texas

How does it know what "look" you are trying to achieve?

Canisdirus

Except it's pointless...

There are better topics...real topics...

Not...DSLR or Mirrorless threads...or...
All you need is SOOC threads...holy smokes...talk about inane conversation.

LOL...all brought to the community from 'The Dead Horse Club'

Longshadow Loc: Audubon, PA, United States

It will tell you what you want.


Oh, wait, that's already being done.

JBRIII

As in many discussions on this site, we often seem to fall into arguments over the use of words. Words are important and I'll admitt we often don't have the correct ones to correctly describe things properly.
In the 70's and 80's 90?, programming tried to mimic human intellegence by explicit programming line by line instructions for how to translate languages, makes things walk, etc., it failed for decades because we often don't know how we do it. It was then also called A.I. and sometimes I believe expert systems. Modern A.I. tries to minic nerve connections in the brain and the computer decides what makes sense. So it is artifical, but limited intelligence (I.).
You can't say natural or human I. is always better since A.I.has beat the best humans at games, chess, go, etc. But each system plays one game and also often relies a lot on being able to look ay many more possible moves than a human can in the same period of time. So it is in a way doing what humans do and is A.I. There are words to describe what type of processing is done, but like most fields of study, they make sense only to those in the field. I'm sure an expert in optics could actually describe "lens compression" in terms completely concise and acturate and meaningless to the rest of us. Just as a chemist could state that H2 + 02 + a spark releases X amount of energy due to valence states and electronegativity differences.

Finally, science tries to describe things and name things so a general audience gets some understanding. At work, all manuscripts also required a non-technical abstract for the lay audience aimed at 8th grade level. Lignin is a non-repeating polymer made from various coniferyl alcohol monomers, meaningless to most, so I called it the woody stuff in plants. Since lignin does give wood it's strength and is the hardest part of plants to digest, the meaning was good, but not accurate as wood contains cellulose, etc. and lignin is in all? plants, (trees, bushes, grasses, alfalfa, etc.) Global warming was easy to understand, but climate change is more accurate, etc.

I always believed that somewhere the verbage used in the abstracts would cause problems, somethings just can not be accurately described in simple terms in a limited space. Today with the net everything written is there for people to see (good) and pick apart to feed their conspiracies (not good). Something needs to be done to better communicate with the public, but I really don't know how without overwhelming people who are already overwhelmed.
Jim

TriX Loc: Raleigh, NC

Good summary JIm, and personally, I prefer expert systems or machine learning to AI.

User ID

So, you have little if any experience with the topic at hand. But thanks for posting so we know you’re still alive.

User ID

There were talking cameras in the ‘70s.
IIRC it was not a blazing sales success.

JBRIII

Thanks;
Jim

TheShoe Loc: Lacey, WA

Maybe extrapolate?

srsincary Loc: Cary, NC

Something as "simple" as auto-eye tracking in dynamic AF mode uses deep learning, which involves training a many-layered artificial neural network (never intended to directly emulate our brains biology BTW) to detect an eye in the image captured on the camera sensor. And keep doing that as the image changes.

Computational photography in the iPhone, Google Pixel, etc. also use deep learning.

So does AlphaGo, the world's best Go player. So does any automated facial recognition system, that the cops love.

Topaz AI SW uses DL to detect noise. DL computations use a lot of matrix operations, which is why DeNoise and Sharpen run slow without GPUs.

DL is a subset of ML (machine learning) is a subset of AI.

IBM's Watson which won Jeopardy in 2011 did not use DL, but did use some ML and definitely a lot of AI. As a long time Big Blue employee (note my signature 🙂), I had the privilege to study some of the code, though I lacked the expertise to be part of the team.

AI (the superset of ML and DL) is all around us, like it or not.
👍🙏

User ID

There is no comparison possible.

Believing that you could compare the two suggests that “actual” human intelligence is sadly lacking.

User ID

“AI” is another of those poorly crafted marketing buzz terms like “mirrorless”.

TriX Loc: Raleigh, NC

Exactly (as I stated two pages ago).

burkphoto Loc: High Point, NC

I remember reading about that in Infoworld. Wow. Long time ago!

TriX Loc: Raleigh, NC

When you get to my age Bill, everything is a long time ago