2019-04-09 at 21:24 #10821Moises TejadaParticipant
Hello Greg. I Think I came up with an idea to implement the “tracking in 360” degrees for Kinect.
I notice you do not know anything about programming or the operation of the Kinect software in Windows with your SDK, but I think this idea (Which may be idiotic) can work.
My idea would be to have 2 kinects in a way and at the same height (i Think the distance between them would influence and should be some distance in particular)
These Kinect would be guided by 2 different colored balls put in our heads, using the color tracking over our head, each color ball on the front and back of our head, something like show in the image
(Could be Unlike but it is what I think that aumentatia the precision of the tracking of rotation in our body and know that Kinect is looking at the Kinect)
Developing The idea would be like this. The Kinects only Detectarian the body in front (as I know) and when we pass the 180 of rotation with our body and emperzaria detect it in a errornea way and decalibrate. I thought that with the two colored balls using the 2 Kinect connected you could “scan” the body.
Each color would be associated with a Kinect, Example: The Kinect 1 to the blue color and the Kinect 2 to the red color.
With This association would be created as a follow up of our body and its turn which would make that when we 180 degrees in front of a Kinect these 2 balls would be visible in the 2 Kinect (Here Acentuo that as is obvious when you turn your head but not the body these 2 ball s would be visible in the 2 Kinect at the same time which would come into play the fact that the Kinect that scans us in front could calculate our body entered and its turn, including the tracking of the colors) The turn of our body would be an essential part to start power off Tivar a Kinect and activate another so create that “Space 360 º”.
You could ask “what happens when we are sideways in the 2 Kinect and we do not look at any front?”
Here comes the most specific function of the colored balls. Having the 2 in the head and one in our forehead and another in the back of the head and when the 2 Kinect will not be able to calculate our body correctly because we would be on the side of the 2. Just these balls Eliseo to which side we would be respectively from each Kinect.
Example: We Are Looking to one side that is not to any Kinect. Each Kinect detects only half of our body and to know that half is and to which side we are looking because the color tracking would be in the two Kinect and the kinects or the SDK could “Put Together” each side of the body (or skeleton) that is scanned and join it virtually. And as from the sight of each Kinect the position of each color balls would be different
And when you look at a Kinect or detect enough body or limbs to work with 1 single Kinect, the colouring would stay in wait and it will activate again when this same Kinect detects a turn of more than 180 degrees and would return to activate the 2 Kinect for Do a scan of each side of body and join it in the “virtual skeleton” and with the help of the orientation of our position with the ball of colors could Also ask “What if we have a side position on the kinects and maintain that position while GI We head to the other side, losing the color tracking and would lose the position on the Kinect and could not make the virtual union ”
As This union is to be found until the 2 Kinect that our body is turned and know that color is front or back.
I Do Not know if this idea is stupid but I wanted to tell you because I think a very good job what you’re doing.
I’M Sorry If you don’t understand some part since I translated it with the Google translator.
A greeting2019-04-10 at 10:54 #10823Greg DriverKeymaster
thanks for this documentation. i think i will start with something simple – that is to use skeleton tracking and 360 degree rotation. then we can go to the colour tracking.
it is no problem if two kinects see the ball cause they can then use info from both even better given there is calibration done.
however for head I would place ball on the top. then you have single ball.2019-04-19 at 23:46 #10889ZihaoParticipant
Damn! that’s some cool idea. I know like C,C++ stuff but nothing about the Kinect SDK. Where do you think I should start learning that?2019-06-02 at 15:07 #11412Greg DriverKeymaster
get sdk samples and run them they are not big 🙂2020-06-24 at 19:45 #15348JuliLeThingParticipant
This is a good concept for the 360 tracking but I don’t think that only 2 kinects would be enough since the kinect only has tracking of 90 degrees, so if you want to have 4 kinects. 1 in front of you, 2 on your left and right and one behind you.
- You must be logged in to reply to this topic.