Skip to content

superanimal_humanbody with FilteredDetector: add video adaptation#3086

Merged
AlexEMG merged 66 commits intomainfrom
maxim/superanimal_humanbody_video_adaptation
Sep 10, 2025
Merged

superanimal_humanbody with FilteredDetector: add video adaptation#3086
AlexEMG merged 66 commits intomainfrom
maxim/superanimal_humanbody_video_adaptation

Conversation

@maximpavliv
Copy link
Copy Markdown
Contributor

@maximpavliv maximpavliv commented Sep 3, 2025

This Pull Request extends #3066, by adding the possibility to run video adaptation with the superanimal_humanbody pose model.

Additional notes:

  • The detector adaptation is not performed for superanimal_humanbody since the FilteredDetector is used (rather than a native DLC detector). A message warns the user about this.

  • A NotImplementedError is raised in build_weight_init() if super_animal == "superanimal_humanbody", to prevent attempts of TransferLearning and Finetuning from superanimal_humanbody weights. This feature could potentially be implemented in the future.

@maximpavliv maximpavliv changed the title superanimal_humanbody with filtered detector: add video adaptation superanimal_humanbody with FilteredDetector: add video adaptation Sep 3, 2025
@maximpavliv maximpavliv force-pushed the maxim/superanimal_humanbody_video_adaptation branch from 86171bc to e9ff1a1 Compare September 5, 2025 18:21
@maximpavliv maximpavliv requested a review from AlexEMG September 10, 2025 13:48
@arashsm79
Copy link
Copy Markdown
Contributor

Just pulled this and managed to run superanimal_humanbody with/without adaptation through the GUI and the API.
LGTM!

@AlexEMG AlexEMG merged commit 10e9beb into main Sep 10, 2025
5 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants