For now, I don’t need to find the point in real time, I can cost 2~3 second to calculate the two points( only needs Y-coordination )
So my need is pretty simple, But what I found on the internet is find the 2. chin only, but for the top of head… I haven’t found a good post yet.
What I have found is this post
So somehow it can extract the hair, so I just get the topmost point of hair segmentation as top of head, then I can figure out these two point.
But I just wonder, does this simple request needed to be commit so much efforts? are there simple ways to do?
And since hair is varied a lot, maybe the person is bald or with colorful hair, but usually is black, white or gray
I need to consider all of these cases.
i would not think you need something as complicated as hair detection.
opencv comes with several face detection methods, like cascades and dnn based detection, there are also facial landmark detectors, which can give you the actual face outline.
I check topmost results of google “facial landmark detectors” and “facial landmark detection”, but what I have seen, they don’t detect “top of head”(need to consider hair), they do detect chin, and face outline.
For example, I found this one post
it says if you want detection area include hair, then you need to train your data, since current model not include that.
If there has something I have missed, please tell me.
We need to extract head for printing photo, let’s what we want to do.
But I found a way to solve that using Canny to detect edge will probably solve my problem