I need to implement people counting(head count) into one of the product. I am able to detect faces and count them but only when the persons are actually looking into the camera, if they are not looking into it they will not be detected.

So, what I want is to an idea about detecting human rather than detecting only face.

I have gone through some algorithm like voila jones etc..but did not get anything useful.

It would be great if someone can share their experience or any code snippet to do so or any algorithm to help me out here.

  • $\begingroup$ Did you find any way to do this? I also get stuck in same issue. $\endgroup$
    – Srikanth
    Commented Jul 4, 2018 at 6:33
  • $\begingroup$ I am planning to work on the problem. Came across this research paper. They have proposed a method to perform image crowd counting using CNN and MRF. Looks promising. $\endgroup$ Commented Oct 13, 2018 at 6:49

1 Answer 1


This is an ongoing research problem, which is much more difficult to solve than face detection due to a gigantic variety of poses, angles, and other factors that play a role. This is usually done with image feature extraction on a large set of images followed by a supervised learning algorithm. In short, there's no unique answer, and there's definitely not a simple "code snippet" one could use for this task.

If you're implementing a specific algorithm, you could probably get better help here, but not at this point.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.