search

google-ai-edge / mediapipe

Cross-platform, customizable ML solutions for live and streaming media.
https://mediapipe.dev
Apache License 2.0
26.62k stars 5.07k forks source link

Is it possible to obtain accurate 3D world coordinates for each keypoint relative to the camera by providing depth images? #5062

Open CGuangyan-BIT opened 7 months ago

CGuangyan-BIT commented 7 months ago

MediaPipe Solution (you are using)

directly utilize the resulting world coordinate

Programming language

Python

Are you willing to contribute it

None

Describe the feature and the current behaviour/state

The accurate 3D world coordinates according the depth images instead of the estimated depth

Will this change the current API? How?

No

Who will benefit with this feature?

Yes

Please specify the use cases for this feature

Accurately obtaining the position and orientation of the hand is required

Any Other info

No response

kuaashish commented 7 months ago

Hi @yichunk,

Could you please have a look into this issue?

Thank you!!

Thaina commented 5 months ago

I have depth camera and I wish I could provide depth data for accurate 3D landmark like kinect (in fact it is kinect and want to use mediapipe with it)