2
$\begingroup$

At this moment, I am able to use NN to identify object such as human when given a frame from the camera. Once locate the object, then I can feed the human object image to either NN that's designed to classify male or female.

Let's say I get 1 frame per second from camera and perform detection, the objective is to track number of male and female walk pass the camera within the given hours.

My question is, the same person in multiple frames will be over counted. I couldn't wrap my head around how can I train a NN to understand that this is the same person without dive into facial recognition? I'm sure there is some tracking technique that I just don't know.

One little constraint, if the person left the camera frame and come back into it later, it is fine to treat it as two people.

Any help or direction will help!

Thank you all in advanced!

$\endgroup$
  • $\begingroup$ Welcome to AI! Was wondering if you might be able to provide an answer on this question, related to identifying human beings with NNs: ai.stackexchange.com/questions/4678/… $\endgroup$ – DukeZhou Dec 9 '17 at 23:24
  • 1
    $\begingroup$ I went there and attempted to leave a comment, but didn't really work because of need 50 rep. Alternatively, I left it in the answer instead. Hope it'll help! $\endgroup$ – WorldWind Dec 10 '17 at 6:06
  • $\begingroup$ Thanks for that! I think it's a suitable answer re: how NNs work in relation to the problem, well suited for an OP without any background in the methods. (The question had been edited from it's original form, which was much more general.) $\endgroup$ – DukeZhou Dec 10 '17 at 21:06
1
$\begingroup$

One way to solve the issue may be a tracker (like Kalman). It would be faster and easier approach than neural nets.

If you insist about solving this issue through neural nets, then some magic and creativity is needed. Based on the nature of tracking, you need to feed multiple frames to predict next location of the object and check if there is any object close by. So, you may consider combining RNNs with CNNs to predict next bounding box to track and replace Kalman filter's prediction with RNNs. (check)

| improve this answer | |
$\endgroup$
  • $\begingroup$ Thanks for the quick reply. So, is there no real tracking? I believe Kalman filter is using a predictive motion model. So, if I only have 2 frames, I cannot track the object. Also, what if the object is human? I cannot really predict the motion as human may stop and turn around at random time. Any insight to my situations? $\endgroup$ – WorldWind Dec 13 '17 at 7:32
  • $\begingroup$ It would be better for you to check how tracking algorithms work. Tracking algorithms are designed to solve those kinds of problems. $\endgroup$ – Deniz Beker Dec 13 '17 at 13:04

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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