Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

example for single pose lightning model use #16

Closed
wants to merge 4 commits into from

Conversation

stephanschulz
Copy link

As mentioned here I noticed that the current movenet example does not work well when a person is close to the camera.
So, I made this example using the single pose lighting model.

#11 (comment)

I don't do PRs very often and hope it did not make any mistakes.

The current exaple_movenet sues the multipose lighting model which does not work well when a person is close to the camera.
https://tfhub.dev/google/movenet/singlepose/lightning/4

This is an example using the singlepose lighting model:
https://tfhub.dev/google/movenet/multipose/lightning/1
@danomatika
Copy link
Member

Is the model included? If not, that is fine as we host them separately since they would make the repo much larger. Can you provide a link the model?

@danomatika danomatika added the feature New feature or request label Feb 9, 2022
@stephanschulz
Copy link
Author

I got the model from here: https://tfhub.dev/google/movenet/singlepose/lightning/4
and downloaded TF2.0 Saved Model (v4)
I see now that there are other options/tabs too like A TF Lite format of google/movenet/singlepose/lightning/3 but I did not try those.

@danomatika
Copy link
Member

danomatika commented Feb 9, 2022

Ok, thanks.

Our feeling is that we don't intend to gather all available models into examples within this repo. This example is basically similar to the existing movenet example, so it might make more sense to house it elsewhere.

At the same time, I think it is definitely useful to add info to the existing movement example readme of how you adapted the current example to run a variant of the original model, ie. looked at model inputs/outputs and adjusted the code for it to work.

@stephanschulz
Copy link
Author

sounds good.
thanks for all your work on this.

@danomatika
Copy link
Member

Closing for now.

@danomatika danomatika closed this Mar 31, 2022
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
feature New feature or request
Projects
None yet
Development

Successfully merging this pull request may close these issues.

None yet

2 participants