Recently, I have being planning to design an automatic pornographic movie detector with the machine learning knowledge that I just acquired.

I am now in the stage of planning how to perform the system evaluation. The intuition is to just test the algorithms with some pornographic movies and some non-pornographic movies and compute the false negative/positive rates. I simply feel that for any evaluation method, I would have to obtain real pornographic movies to test the system!

Here comes the dilemma. As you know (or may not know!), basically all the pornographic movies are not free. I do not regard it as ethical to just download illegal pirated movies, as after all they are for a scientific research, which values copyright a lot. On the other hand, it is a bit awkward for me to request funding from my PI to buy the porn videos! Paying them myself is definitely not a feasible solution, either.

How may I handle this case properly?

  • 7
    Maybe you can add this paper to related work: "Detecting Obscene Content and Misbehaving Users in Online Video Chat Services": a system for figuring out when someone on, e.g., Chatroulette is habilimentally challenged. Was sitting in a random session of WWW 2011 when this talk was given out of the blue. Was a highlight of the conference.
    – badroit
    Commented Apr 7, 2014 at 23:27
  • I'm not familiar with the applications of this kind of work. Would it be possible to add a brief explanation or a link explaining possible use cases? Commented Oct 28, 2014 at 6:10
  • This is going to need a lot of country specific tuning; French daytime TV used to include nudity that would not have been allowed in the UK late at night. Then add in religeous and political variables.
    – Phil Lello
    Commented Mar 30, 2016 at 21:29

4 Answers 4


I really don't get this. This is a little like the Movies Classification Board or the Classification & Ratings Administration feeling awkward about asking to view the movies they are going to rate.

If you are developing an algorithm for detecting pornography, why would it be awkward to request funding to acquire test resources? This is an entirely legitimate application, that has huge potential, both economically and socially (note: use of bold inspired by OP). There are many researchers all over the world working in similar fields. Clearly, at some point you will need to discuss your work with your PI, and presumably, they are already aware of the work you have been doing in the previous 6/9/18 months, and perhaps already requires justification against existing research funding.

Whatever the case, your work will need to be approved by, and comply rigidly with, your institution's ethics committee.

There are also a number of steps you could consider before committing to viewing full-length extreme hardcore films. In making some suggestions, I'm going to make a few assumptions:

  1. 1 Your definition of a pornographic movie is one classified as "mainly concerned with sex", and involves extreme nudity and acts of sexual behaviour.

  2. Your algorithm looks at exposed and interacting body content in some way, and does not work on quality of script and acting (otherwise daytime soapies will bring up many false positives)

  3. Your algorithm looks at small fragments of movie clips, and not the entire 60/70/90 minutes worth, or however long these movies go for.

  4. Your PI knows what you've been spending your previous doing, which is most likely already associated with existing research funding.

So, some suggestions

  • Acquire a number of movie trailers; these are usually freely available for all types of movies, and typically run for about a minute. Contact the various movie studios directly if you need to.
  • Speak to your national Film and Television archive body. They are likely to have a number of suitable media to preliminary early test procedures on. Various bodies exist.
  • Speak to other researchers in this field. There might be a good standard set of images/movies that can be used to determine baseline effectiveness.
  • Go to your local friendly adult shop and check out the "specials" bin.
  • Finally, and not the best approach to take, look at the internet. I'm guessing if you typed in the right combination of words, you'd get unlimited supply of short, copyright-free videos, to use both as a test group and control group. But I stress, this needs to be done in strict accordance with your institutions policies for network and computing use, and therefore should be approved by your PI and ethics committee.

However, whatever approach you take, the summary is: talk to your PI.

  • Thanks for the detailed answer. I think I should just talk to my PI. One thing that stops me from requiring funding from my PI is that I feel it not worthy it to buy the movies. Like others have mentioned, one movie is usually 30-min long maybe. I don't really need it to be that long. So I feel paying the full price for it with the funding is kinda waste Commented Apr 8, 2014 at 15:35
  • "spending your previous doing". Missing word? Commented Oct 28, 2014 at 6:07
  • The answer is right that you can get for free a lot of clips from the Internet (that may be even more true about any other film gender), but they aren't likely to be copyright-free unless you are interested in 100 year old vintage porn. However, if you are just analyzing the films and not releasing them, copyright isn't likely to be a problem.
    – Pere
    Commented Jul 6, 2018 at 10:04

As you know (or may not know!), basically all the pornographic movies are not free.

I find this not quite realistic: A fair chunk of the total internet bandwidth goes to porn. There are a lot of "free" websites. And a lot of "studios" offer samples which would suit your research, in my opinion.

You don't need a two-hour porn movie to test your algorithms.

Furthermore, what is the state of the art in this field? Is anybody else working in that domain? What are they doing, i.e. how are they getting films to test?

  • 9
    Some nontrivial portion of those "free" websites are copyright violations. Though this might be one of the few cases where a legitimate Fair Use argument could be made.
    – keshlam
    Commented Apr 8, 2014 at 3:24
  • 11
    +1 for "You don't need a two-hour porn movie to test your algorithms."
    – Suresh
    Commented Apr 8, 2014 at 6:44
  • 20
    Who needs a two-hour porn movie for anything? Commented Apr 8, 2014 at 11:05
  • 1
    They simply do not mention how they obtain the videos. Who knows they download it or buy it. :) Commented Apr 8, 2014 at 15:40
  • 1
    @MarcksThomas, Hmm... Tantics? :)
    – Brian S
    Commented Apr 9, 2014 at 15:39

If your supervisor knows that she is funding you to do research about pornography, I don't think she should be particularly shocked by the the request. If your supervisor does not know, that will be the more awkward conversation and it sounds like its time to have it.

Additionally, good options might be:

  • Look for freely licensed pornography. Creative Commons provides a way to search for freely licensed material including a way to do Google searches for permissively licensed video.
  • Try to rely on a fair use justification and go ahead and use copyrighted works you can find for free online (e.g., from free porn video websites).
  • You could ask commercial pornography distributors to donate material for science.

Of course, there are risks with going with the fair use argument and it's always best to get your institution's general counsel's opinion first.

  • 1
    It's probably a good idea to state (emphatically) several times that the material is not going to just be used for masturbation.
    – Brad Werth
    Commented Apr 8, 2014 at 6:47
  • 8
    Not "just".
    – tobyink
    Commented Apr 8, 2014 at 9:35
  • 1
    Thank you so much on the CC advice! our works currently involve only the generic algorithms for now. i.e., not specific movie type. It is me who just come up with the idea of doing the porn type. So the PI has not approved this "sub-project" yet, strictly speaking. Commented Apr 8, 2014 at 15:40

I believe you're facing a no-problem kind of problem.

Many of the largest porn streaming sites do actually pay content creators so they allow them to serve the videos for free. What you describe is a fair use case, so, in theory, you shouldn't face any issues. Most studio-provided content on these sites are actually legally served for free.

When using one of those sites, make sure you search the studio pages rather than the whole site. Searching the whole site can indeed return pirated results.

(I'm posting as a community wiki because I think this is neither an academic answer nor an Academia.SE-worthy answer. It's better as a comment, but it doesn't fit as a comment)

You must log in to answer this question.

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