Hello everyone:
We want to detect if the people wear a helmet on a video sequence,
as the attached image shown. But it looks there is not a good solution for this.
Could anyone give me some suggestion?
Thanks Yanbin Yue
That's not an easy task.
First of all you want to detect an object that (as shown in figure) hasn't much texture. Detecting such objects is somewhat hard for the current state-of-art object detections method such as SIFT etc.
If your helemts all have distinctive colors (yellow can be good) I suggest you can do soemthing like this:
If the person is close to the camera and the helmet has a distinct color like in your example image i think you can skip the face detection step described above and do just the color detection. Otherwise if the person is far from the camera and there are even multiple persons within the image, doing the face detection and the color detection is a good approach.
Thanks for your answers. We have talked about this solution several days. We think this solution may not be very perfect, because we need solve some other problem in this solution, such as the hsv value of the same color would have much difference in different light intensity, and so on. So we thought that we need some more measures combine with face detect to decide the detect result.
Asked: 2012-08-27 04:52:59 -0600
Seen: 3,224 times
Last updated: Sep 27 '12
You can throw a stone and detect the face reaction using an emotion detection algorithm. Those without helmet should have a strong emotional response.