About the principle of azure kinect dk.

aa 16

I would like to know the principle behind Azure Kinect DK's ability to acquire the position of each human joint.

António Sérgio Azevedo 7,671 Reputation points Microsoft Employee

2021-11-22T23:27:42.977+00:00
Hello @aa please confirm you have looked at the following documentation already, can you be more specific on what you are trying to achieve?

Azure Kinect body tracking joints

Thank you!
aa 16 Reputation points

2021-11-23T12:53:04.537+00:00

I'm sorry for the lack of explanation.
I would like to know the flow from shooting a human with Kinect to outputting the joint position.
António Sérgio Azevedo 7,671 Reputation points Microsoft Employee

2021-11-24T12:15:01.147+00:00

I am leveraging our SMEs to help here. Thanks for your time so far!

1 answer

Quentin Miller 351 Reputation points

2021-11-24T18:26:32.007+00:00
Please sign in to rate this answer.

1 person found this answer helpful.
Quentin Miller 351 Reputation points

2021-11-24T18:30:10.09+00:00

CNN = Convolution Neural Network
The CNN returns up to 32 joint locations per detected skeleton.
The (human) Model Fitter combines the CNN output with depth data, information on the human form, and temporal information to compute the final position of all 32 joints in 3D per skeleton. It will estimate joints missed by the CNN. You can use the confidence level to determine if the Model Fitter has estimated a joint position.

aa 16 Reputation points

2021-11-26T10:44:47.267+00:00

This is the information I want to know! thank you!

António Sérgio Azevedo 7,671 Reputation points Microsoft Employee

2021-11-26T16:10:21.79+00:00

@aa glad we were able to help :). Can you please verify the above answer?

Thank you so much.

Remember:

Please accept an answer if correct. Original posters help the community find answers faster by identifying the correct answer. Here is [how][10].
Sign in to comment

Use comments to ask for clarification, additional information, or improvements to the question.

Your answer