mediapipe

mediapipe copied to clipboard

visibility and presence fields in a NormalizedLandmark for the Holistic Solution always be 0. Is this unsupported right now?

Hi @endink, Could you please elaborate your query with complete details. Thank you!

Sorry, visibility and presence the fields is in MediapipeUnityPlugin project.

Simply, if the the score or visibility of the hand landmark in this Holistic scene available now ? ( like the pose landmarks).

BTW, please forgive me for my English grammar errors.

@endink - We opened the same issue but never got a reply on the reason why it never detects anything other than 0, nor that it was a defect

@jimb2834 we can only wait for a reply...

@kuaashish and @ivan-grishchenko, is there any update on this issue? This issue is affecting other validations such visibility and presence of various points of interest as it is always 0 hence wrong results. One example is palm validation, checks for palm validation and visibility on an image with only points 8,12,16,20,7,11,15,19,6,10,14,19,5,9,13,17, mediapipe is returning other points not available on the image hence wrong result. find sample image below

@elYeodeol - there are old issues related to Visibility and Presence of landmarks re: face mesh specifically from a while back - https://github.com/google/mediapipe/issues/3200 https://github.com/google/mediapipe/issues/3159

But anytime someone solves it, ping me and we are happy to pay for the solution.

@elYeodeol - there are old issues related to Visibility and Presence of landmarks re: face mesh specifically from a while back - #3200 #3159

But anytime someone solves it, ping me and we are happy to pay for the solution.

Hi jimb2834, have you looked at Google mlkit library for face landmarks? https://developers.google.com/ml-kit my experience has been impressive.

@ajmakoni - Hi, thanks for the message. The main issue was landmark quantity; whether it reflects "visibility" or "presence" I did not check. So the landmark Qty, Visibility, and Presence values were all needed for our use case. GCP ML kit for face has a lower landmark qty.

@ajmakoni - Hi, thanks for the message. The main issue was landmark quantity; whether it reflects "visibility" or "presence" I did not check. So the landmark Qty, Visibility, and Presence values were all needed for our use case. GCP ML kit for face has a lower landmark qty.

Okay, I get your point

Is there any good news for this? Nobody care it?

@endink - I was in the MediaPipe slack channel and saw this

@jimb2834 it only worked for pose landmarks, but not in hand landmarks.

Hello @endink, We are upgrading the MediaPipe Legacy Solutions to new MediaPipe solutions However, the libraries, documentation, and source code for all the MediapPipe Legacy Solutions will continue to be available in our GitHub repository and through library distribution services, such as Maven and NPM.

You can continue to use those legacy solutions in your applications if you choose. Though, we would request you to check new MediaPipe solutions which can help you more easily build and customize ML solutions for your applications. These new solutions will provide a superset of capabilities available in the legacy solutions. Thank you

This issue has been marked stale because it has no recent activity since 7 days. It will be closed if no further activity occurs. Thank you.

This issue was closed due to lack of activity after being marked stale for past 7 days.

Are you satisfied with the resolution of your issue? Yes No

Mediapipe unity plugin stills gives always 0 for visibility and presence for normalizedlandmarks in handtracking