Skip to content

RTMpose_x superanimal_humanbody#3043

Merged
AlexEMG merged 35 commits intomainfrom
mwm/humanbody
Jul 6, 2025
Merged

RTMpose_x superanimal_humanbody#3043
AlexEMG merged 35 commits intomainfrom
mwm/humanbody

Conversation

@MMathisLab
Copy link
Copy Markdown
Member

@MMathisLab MMathisLab commented Jul 5, 2025

Major Changes:

  • This integrates the RTMPose model that we demo with the low-level API in this notebook fully into the model zoo.

  • For ease, this is a new function called deeplabcut/pose_estimation_pytorch/modelzoo/superanimal_humanbody_video_inference.py

  • it also works in the api call

  • It also updates the model-zoo tab for better handling of thresholds and more elegant layout:

Screenshot 2025-07-05 at 13 12 26
  • It also now tells the user if not all frames have a detected animal:
Screenshot 2025-07-05 at 13 13 18
  • Right now it does not support explicitly support video adaptation in the gui.

Minor:

  • fixes to some missing info on superbird

@MMathisLab MMathisLab requested a review from Copilot July 5, 2025 09:09

This comment was marked as outdated.

- adding missing docs for superbird
@MMathisLab MMathisLab requested a review from Copilot July 5, 2025 09:18

This comment was marked as outdated.

@MMathisLab MMathisLab requested a review from Copilot July 5, 2025 11:24

This comment was marked as outdated.

@MMathisLab MMathisLab requested review from AlexEMG and Copilot July 5, 2025 14:35
Copy link
Copy Markdown
Contributor

Copilot AI left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Pull Request Overview

Integrate the RTMPose_x human‐body model into the DeepLabCut model zoo, add a dedicated video inference pipeline using torchvision detectors, update configs, docs, tests, and GUI layout.

  • Added superanimal_humanbody_video_inference.py for a clean, Colab-style human‐body demo
  • Extended load_super_animal_config and utils to download configs from HuggingFace and support torchvision detectors
  • Updated tests, documentation, and the GUI model-zoo tab for thresholds, layout, and human‐body support

Reviewed Changes

Copilot reviewed 24 out of 24 changed files in this pull request and generated 1 comment.

Show a summary per file
File Description
tests/pose_estimation_pytorch/modelzoo/test_webapp.py Include superanimal_humanbody in webapp tests
tests/pose_estimation_pytorch/modelzoo/test_modelzoo_utils.py Include superanimal_humanbody in utils tests
test_superanimal_humanbody.py New top-level script for human-body detector integration test
deeplabcut/pose_estimation_pytorch/modelzoo/utils.py Add HF download and special handling for human-body configs
deeplabcut/pose_estimation_pytorch/runners/inference.py Introduce TorchvisionDetectorInferenceRunner
deeplabcut/gui/tabs/modelzoo.py Revamp GUI: thresholds, media selector, adaptation options
Comments suppressed due to low confidence (2)

deeplabcut/pose_estimation_pytorch/runners/inference.py:866

  • BaseDetector is not imported in this module, causing a NameError. Add from deeplabcut.pose_estimation_pytorch.models.detectors.base import BaseDetector at the top of the file.
class TorchvisionDetectorInferenceRunner(InferenceRunner[BaseDetector]):

deeplabcut/gui/tabs/modelzoo.py:365

  • deeplabcut.superanimal_analyze_images is not exposed on the deeplabcut module at top level, leading to an AttributeError. Import or fully qualify this function before use.
            help_text = deeplabcut.superanimal_analyze_images.__doc__

Copy link
Copy Markdown
Member

@AlexEMG AlexEMG left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Excellent

@AlexEMG AlexEMG merged commit ecc06f8 into main Jul 6, 2025
4 of 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.

4 participants