Computer Vision is Artificial Intelligence

I grew up reading Asimov's robot stories and knew I wanted to work on artificial intelligence growing up. I realized when I got to college that computer vision was the closest thing to what I had always thought AI research would be like. (So I did a PhD in computer vision.)

On a related note, David Lowe, of SIFT fame, once related to me a story about when Isaac Asimov visited his research lab in the early 1980s and when they explained to him that computer vision was about enabling robots to see the world, Asimov shrugged and said he never realized that was even a problem -- he just assumed robots would be able to see.

If anyone wants to work on "artificial intelligence," then, GazeHawk is hiring!

(sorry for the spammy post, but I'm really impressed by the level of computer vision experience and talent showing up in this thread, and wanted to reach out to you all. We are a startup which is also actually solving hard CV problems)

I quit my Phd in computer vision. There where i was working with a stereo vision robot. So i was trying to get the robot to behave like a human would, fixating point of interest (salient points) and calculating distances, well that involved a lot of trigonometry and the cameras had to be perfectly calibrated to get good data, so it was really hard just to match points of interest from both cameras. In conclusion i believe robots will see the world, but it will be through their eyes/sensors, not some human like approach. As for AI it will have to be some sort of baby like bootstrapping aproach able to learn, validate, propose new theorems and have it's own curiosity, in the form of a function that minimizes entropy or something like that.

And it will be funny if humans won't be able to make sense of that learning data :)

It's certainly no less AI than natural language processing is -- both seek to use algorithmic approaches to resolve the ambiguities in a bunch of input captured from fuzzy real-world sources, and build data structures representing that data -- as best understood by the program -- that can be manipulated by more traditional processing algorithms.

I have a degree in Machine Learning. Recently I started working on a smartphone app involving CV. I am working closely with CV researcher.

And guess what? They use vectors, linear algebra, probability and classifiers too! They just go to different conferences to talk about them. If you learn enough math you can see through most of the terminology differences.

Those of us in the so-called real world must never forget the harsh industry structure of academia rewards fragmentation and obscurity, not unification and accessibility. It's the fault of the customers (grant makers) for rewarding this labor maximization behavior.

Deep Blue was "in the early days of AI"??? The OP needs to study the field before making such pronouncements. There were many breakthroughs in AI 30 years before Deep Blue. See, for example, Arthur Samuel's checker program from the early 1960s. http://en.wikipedia.org/wiki/Arthur_Samuel

My AI lecturer remarked that AI was a great term for hard problems, and that once a field or area was effectively solved or developed, it no longer got called AI.

"Artificial Intelligence" is both an umbrella term for a number of CS/stats approaches to certain types of problems (graph search, classification, etc.), and also a term often used to mean the same thing as Artificial General Intelligence (i.e. an artificial mind).

Vision is trivially covered by the former definition, and clearly not by the latter.

As a researcher working in computer vision, personally I prefer not having computer vision associated too much with AI just because of the pejorative aspects that the term artificial intelligence has had in the past.

Nice little, makes me think...

"I believe that once we have made progress on vision (not in the narrow-universe setting) to the point where generic visual scene understanding is effectively solved, there won't be much left that needs to go into the "ethereal" mind which cognitive scientists want to empower machines with..."

This is a great point but it may have implications that won't make the author happy.

If the field of vision computer vision is "nearly as big as" the field of general artificial intelligence, it may not be a good direction from which to approach general artificial intelligence.

The vision problem involves low-level problems of light-physics-etc, mid-level problems of object transformation and distortion and high level problems of cognition. Saying that vision is hard is saying we need to find one algorithm which can fluently mix these levels rather than using several algorithms that just sequentially deal with each level. And such a level-mixing algorithm sure sounds close to general IA.

It fits with argument of people like Jeff Hawking that there's a "Kernel" of general AI that computer vision is just an instance of (see http://books.google.com/books?id=Qg2dmntfxmQC&lpg=PA101&...). But there's nothing there that says computer vision is useful direction to approach that Kernel from - the technical considerations of vision are hard and you haven't shown it provides a particular useful division in a divide and conquer strategy.

You can experience this first hand if you have ever played or hacked with Kinect.