Vocabulary: Eye AF and Related Terms

Eye-Start: A system invented by Minolta (?) whereby the camera turned on and began to autofocus when you put your eye to the viewfinder. Now used in some cameras to switch between the rear viewing screen and the eyepiece viewfinder.

Eye-Controlled Focus: A system invented by Canon for film SLRs in which a series of sensors in the viewfinder eyepiece read the "bump" of your cornea to cause the camera to autofocus on the area your eye was looking at in the finder. Successful implementation varied between individuals but for many people it worked well. Available on a succession of cameras between 1992 and 2004, the technology did not make the transition to digital.

Face Detection: A digital camera feature whereby the camera recognizes human faces and uses this knowledge to set the autofocus point, usually, but not necessarily always, on the nearest face.

Eye Autofocus (usually written eye AF): A refinement and further development of face detection whereby the camera recognizes the eyes of the subject and focuses on them. Current eye AF has gotten so good that cameras can prefer the nearest eye of a face, and, in continuous AF mode, track the eye and keep focus on it as the subject moves.

Face (properly facial) Recognition: "a biometric software application capable of uniquely identifying or verifying a person by comparing and analyzing patterns based on the person's facial contours. Facial recognition is mostly used for security purposes, though there is increasing interest in other areas of use." (Techopedia)

Mike

Original contents copyright 2019 by Michael C. Johnston and/or the bylined author. All Rights Reserved. Links in this post may be to our affiliates; sales through affiliate links may benefit this site.

(Not) everything must fade away

(To see all the comments, click on the "Comments" link below.)
Featured Comments from:

Terence Morrissey: "As a long time Nikon user i recently purchased the Sony A7RIII for the EyeAF feature and it rocks!"

Herman Krieger: "A really intelligent camera—sees a scene, and decides that it is not worthwhile to take a photo."

Mike replies: That feature used to exist. It was called, "each exposure costs you XX¢ and X minutes at the developing sink."

Mike Schwarts: "If I could afford a new camera I would get the Sony a6400. It combines eye and face detection with a really good focus tacking feature. I believe it can be set to eye detect as default and jump back and forth as it loses the eye and/face and come back when detecting either again."

William Schneider: "I really appreciate the facial recognition feature in Lightroom software. When scanning my father's Kodachrome slides, it was a huge time saver having software recognize and tag faces in family photos. When exporting them from Lightroom as JPEGs, I included the names in the metadata.

"The feature isn't perfect, i.e., faces change as we age, but it saved a ton of time making the archive."

JOHN GILLOOLY: "The firmware update that augments the continuous eye-tracking on the Sony A9: gamechanger."

John Merlin Williams: "Re: Face Recognition. One of the earliest uses was a CNN/IBM-developed application that could identify newsworthy individuals in live-stream video. I saw this in production in 2000. CNN received simultaneous and continuous satellite feeds from more than 30 news agencies. The system could identify and tag most people in those feeds, on-the-fly, from current and past stories, previously identified by CNN editors. They could capture segments of just the pertinent people from live feeds, and do almost instant parallel searches of previous clips, providing a video dashboard of options to an editor. Example: watch for the faces of the U.S. President and U.K. Prime Minister seen together in a live feed and present the editor with multiple clip options to yield: 'After yesterday's meeting (old clip), the two leaders announced this morning (new clip)....'

"It's hard to describe how magical this was (almost 20 years ago); all this could take place in a few seconds. CNN used this to deliver edited stories before other broadcasters, establishing CNN as the first place to go for breaking news. They had a similar system for voice recognition to identify the speaker(s)—in case the editor wanted only feeds and clips in which both persons spoke in the same segment (not just one). I was in awe then, and sad to see that face recognition is most often cited in stories about surveillance/privacy intrusion."

Comments

[Ernie Van Veen]

When I was still a Facebook user I noticed that as I uploaded photos that included the faces of any of my Facebook friends, Facebook instantly and automatically "tagged" them by name on the photo.

[Scott]

"the camera recognizes human faces and uses this knowledge to set the autofocus point"

Sony announced that it is now expanding this to include non-human animal faces (and eyes).

[David Zivic]

Thanks Mike, the last couple posts were very informative as a composite of all of the fantastic new features built into the new and modern, cameras. All that's left for me is to walk around with settings set at M and look for that Decisive Moment. Or do you think Artificial Intelligence will figure that out also.....

[Dennis Ng]

You feed picture and (supervisor type learning) give a label said “steeet photography”, I think the AI can do it if there is enough data (big data). It can even generate it (call GAN). What shock is the self learning one and the AI classify then produce its own genre. Yes, I think AI can do it given the picture and tag on Flickr

For face recognising I do a bit conferences photography it is kind of great to use my Olympic vs my nikon gear as it can do silence as well as face recognising even with em5 II. Just not silence enough for hdr. Need hdr due to the lightening issue. May be Sony better.

[s.wolters]

Big Brother is there, even in our beloved cameras. I always liked face detection. Even works with paintings and sculptures. Told nobody of course because that might undermine my status as serious picture taker.

Here is a recent article that describes the current state of the art.
https://www.cio.com.au/article/659035/casino-facial-recognition-tech-foils-problem-gambler-disguise/

In some places in China, like Shenzhen, the government is already controlling their citizens with advanced face recognition. If you cross the street ignoring the red light or are naughty in other ways you will get penalty points. If you collect too many of them they might deny you some civil rights. You won’t get a visa or things like that.

[T. Edwards]

I never understood why eye controlled focus hasn't jumped to digital.

I had one of the early SLR's to have this technology and it was fantastic. Even many years ago when the technology was much less sophisticated than it would be today.

I mean, I put my camera on continuous focus as my dog runs towards me and you can see the focal points flashing 10 times a second as it tries to interpret the scene and pick out what should be in focus - Is it the shape moving towards us or is it that clump of grass it is moving past?

It seems like it would be so much easier to just track my eye which is locked solidly on the dog running ...

Fortunately, I have zero training or expertise in this area so it will not fall to me to implement these "very simple" ideas ...

[Frank Gorga]

Hmmm... I know that I'm a geezer.

However, I fail to understand why one would want to turn over to a computer what is arguably the most critical aspect of a portrait getting at least one eye in sharp focus.

[Jack Stivers]

I almost never photograph people and could care less if the camera can "track" faces fast or slow.

I find that when the face-finding feature is turned on, the camera goes into over drive to find faces, whether or not there are "real" faces to find. This does prove amusing at times.

Once shadows in the butt end of a fallen tree caught the attention of face finding feature (I thought it was off). It took me a few second to puzzle out why the focus persisted in going there.

Cheers

[David Bennett]

Why should it properly be facial(and not face) recognition? The link mentions that "Facial recognition is also known as face recognition."

But if it should be, then should it not be ocular (and not eye) autofocus?

[Mani Sitaraman]

Eye autofocus?

f/8 and be there.