2
$\begingroup$

This question has generated a good bit of discussion and controversy recently, mostly centered around the original context in which it was placed. After some rather extended discussion in chat, it seems the OP has agreed to a proper overhaul of the question so that it can achieve its original goals, remain on-topic, and stay reasonably inoffensive to a global audience.

Unfortunately, a proper overhaul would render the existing answer fairly irrelevant. So, how should we proceed with regards to the question?

  1. Unlock and overhaul the question in-place. The existing answer would also have to be overhauld by the answerer, or it may even be better off deleted.
  2. Create a new question, based on the proposed overhaul. This way the old Q&A are retained for historical reference, while the re-written question may get a proper answer and provide value to the community.
  3. Don't bother changing anything. Q&A remain locked and unmodified, and we all just move on.

For reference, here's a rough first draft of the new answer based on what was discussed in chat. (Chat discussion ends around here.)


To what degree can a person's age be accurately determined from images alone?

Let's say you have pictures, or a video, of a person. This media includes views of the person's naked body from practically every angle, from a range of about 3-6 feet. This may include some shots of the person's open mouth and/or exposed genital regions.

Assume that the video and/or pictures are the only information you have. Nothing is known of the subject's medical history or family medical history, and neither the subject nor the family are available for interview, direct physical examination, or testing.

Given that subjects' ages may range from 0 to around 32 years, how accurately can their age be determined from these images alone? Is it possible to achieve accuracy to within one year or less for any or all of these possible ages? What sort of factors would one use to perform such a determination?

$\endgroup$
11
  • $\begingroup$ publications.nr.no/directdownload/publications.nr.no/1355995517/… $\endgroup$ – user19679 Dec 19 '15 at 3:33
  • $\begingroup$ scholarpedia.org/article/Facial_Age_Estimation $\endgroup$ – user19679 Dec 19 '15 at 3:41
  • $\begingroup$ google.co.uk/url?sa=t&source=web&rct=j&url=http://… $\endgroup$ – user19679 Dec 19 '15 at 3:47
  • $\begingroup$ google.co.uk/url?sa=t&source=web&rct=j&url=http://… $\endgroup$ – user19679 Dec 19 '15 at 3:48
  • $\begingroup$ I think the links above provide us with a rather firm basis to believe that it may well be possible and it not to an unacceptably inaccurate degree, either! They're quite lengthy documents though. $\endgroup$ – user19679 Dec 19 '15 at 3:50
  • 1
    $\begingroup$ @user19679 The second one you gave provides another interesting use case I hadn't thought of, with regards to immigration. $\endgroup$ – Iszi Dec 19 '15 at 4:18
  • 1
    $\begingroup$ Could someone please summarize the exact nature of the original question and why it was (or might be) so offensive? Why did the (quite obvious) example of child pornography need to be removed? It seems that the three of you who have contributed so far to this meta post were all involved - please help those of us who weren't. Honestly, if I'd have run across a question with this lack of detail on Stack Overflow, for example, I'd close it as "Unclear What You're Asking." The rest of us have no idea what's going on, and the relevant parts need to be collected here in one place. $\endgroup$ – MattDMo Dec 19 '15 at 4:43
  • 1
    $\begingroup$ @MattDMo The original question was much criticized, down-voted, and close-voted for its very explicit inclusion of child porn as a case example. If you want to see it, Revision 1 is still available in the edit history of the original question. Strictly speaking, the entire matter of child pornography is irrelevant to the question of whether or not biology science can help us determine an individual's age within a certain degree of accuracy from visual inspection alone. Further, it brought the question much more into legal territory - which is even broader and less-defined, and less on-topic. $\endgroup$ – Iszi Dec 19 '15 at 4:56
  • 1
    $\begingroup$ @MattDMo Also, what do you see as unclear? I believe the parameters of the task are quite specific. You have images taken of a naked body from 3-6 feet away. Given that your subject may be between 0-32 years of age, how can you use those images alone to more precisely approximate their age - and to what degree of accuracy? What's unclear about that? It may be a bit broad, in which case I certainly agree it may be appropriate to trim the age range a bit on each end (maybe make it 10-27). But how is it unclear? $\endgroup$ – Iszi Dec 19 '15 at 5:11
  • $\begingroup$ Hello, I really appreciate the effort you have went through to help me out, would it be OK if I used the question draft you wrote for me? $\endgroup$ – user19679 Dec 19 '15 at 21:00
  • $\begingroup$ I'm of the opinion that it's not a biology question and should be closed - meta.biology.stackexchange.com/questions/3111/… $\endgroup$ – rg255 Jan 3 '16 at 22:07
3
$\begingroup$

I know that you have put a lot of effort into this question and have tried to help, but I think that the answer is that even the edit you are proposing will result in opinion-based-answers.

I did the following Google scholar search for Human Age Determination and most everything comes up that you would have to physically exam the person. Either look at dental roots, or growth plates, or possibly through blood tests.

None of those would be possible without access to the subject, and at that point, if there was access to the subject, then the video would only be able to represent the age of the person if something like the metadata accurately showed when it was made.

So maybe we can tell a 2 year old from a 4 year old, and we can definitely tell a 2 year old from a ten year old, but we might have a difficulty determining a 16 year old from a 17 year old, or even a 15 year old from an 18 year old. See where I am going with this... At certain stages of life, especially when there is a great degree of change, such as through puberty, not everybody ages in exactly the same way. I remember some boys with full, thick beards at 14 and some who didn't even show the beginning signs of a beard at graduation.

$\endgroup$
5
  • 1
    $\begingroup$ Answers along the lines of "No, it's not possible, and here are some reasons why..." or a more detailed "For age range A, you can get reasonably accurate to within X years. Age range B you can get to within Y years. (etc.) Here's what you'd look for... But you'll never get to within 1 year because..." would still be perfectly valid. Just because the answer happens to be "No" or "You can get what you want, but not as accurate as you want" doesn't mean the question isn't valid or the answers won't be useful. $\endgroup$ – Iszi Dec 19 '15 at 2:05
  • 1
    $\begingroup$ As we discussed in chat, most people are done losing "baby teeth" by a certain age. The presence of those would then at least put a rough cap on the possible age, and the absence would give a rough lower bound. Nowhere near precise age determination, but it's more information than none. Evidence of the progress of puberty can probably put you within a 5-year range for the average population, particularly if you can tell it's just recently started or completed. (I'm not a medical professional, so I could be off - might be less or more.) Still not down to 1 year, but better than no guess. $\endgroup$ – Iszi Dec 19 '15 at 2:10
  • $\begingroup$ @Iszi oddly enough, that likely makes the OPs distinctive case relevant. The issue would be the identification of the gray areas where if the person is 16 or 18 means the difference between the commission of a criminal act and maybe one that, depending on your moral outlook, is questionable, but not criminal. So really the actual question would be, can we tell the age of people in their teens and early twenties visually. And can we accurately make the distinction if they are 18 (whatever the age of consent is) +/- 1 year by visual evidence alone. $\endgroup$ – AMR Dec 19 '15 at 3:18
  • 1
    $\begingroup$ Strictly speaking, the OP's case example does require quite a fine point of distinction. However, the overall generalization proposed here will make the question much less controversial than its original revision. It also serves to broaden the scope so that answers might be useful to a wider audience. Meanwhile, enough detail is preserved in the question that a proper answer will address the OP's needs. $\endgroup$ – Iszi Dec 19 '15 at 4:11
  • 1
    $\begingroup$ As phrased here, the question is essentially "How accurate can we measure age of individuals from 0 to 32 years, only using visual inspection from a few feet away?". It really doesn't matter what the correct answer to this question is - a complete and accurate response will both encompass the OP's interests as well as those of a broader audience. All without the stigma of the original subject matter. $\endgroup$ – Iszi Dec 19 '15 at 4:15

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .