MovementModeler for Education

Turn Everyday Video into
Stick-Figures and Motion Data

MovementModeler is a small, practical iOS app for getting motion data out of real-world video. Any short clip — staged or in the wild — can become a stick-figure visualization and BODY-25 style pose data with embedded metadata, straight from an iPhone.

View on the App Store Python Tools Repo

What MovementModeler Does

MovementModeler is a way to collect and inspect motion data without setting up a lab or wrestling with a full pipeline first. It’s just: phone, video, data.

Motion Data, Anywhere

Record or import video on the phone students already carry. MovementModeler works indoors and outdoors, including in bright sunlight. As with any vision system, stable framing and reasonable video quality help the tracking.

Pocket-sized data collection Works outdoors Real-world footage

Stick-Figure Visualizations

Generate stick-figure, capsule, or sprite-based motion videos that make movement easier to see and discuss. Students can scrub and replay clips for quick visual QC before doing any analysis.

In-class demos Visual QC MP4 export Multiple rig styles

Pose Data for Analysis

Export BODY-25 style pose data per frame and per person. From there, it’s your world — Python, MATLAB, R, or whatever tools you prefer for features, metrics, and models.

BODY-25 style JSON Per-frame data Multi-person

Tag your data before you export New

Label clips with activity type, camera angle, and target subject right in the app. Your JSON arrives downstream with structured metadata — no separate labeling step, no spreadsheet of filenames. Students spend time on analysis, not bookkeeping.

  • Activity label — Pick from a built-in vocabulary (Gait, Sit → Stand, Balance, Dance, Upper Limb, Stairs, etc.) or type a custom label for your specific domain.
  • Camera angle — Tag as frontal, sagittal, or other. Downstream scripts can handle perspective-specific analysis automatically.
  • Subject selection — In multi-person scenes, mark which tracked person is the primary subject. The index embeds directly in the export.
{
  "fps": 15.0,
  "lmm": {
    "activity": "sit_to_stand",
    "subject_index": 0,
    "camera_angle": "sagittal"
  },
  "frames": [ ... ]
}

The lmm block embeds alongside your pose data, so every export is self-describing.

Where It Might Fit

These are just a few places MovementModeler can be useful.

Computer Vision & Machine Learning

Pose Estimation & Beyond

  • Intro labs on keypoints, occlusion, and tracking.
  • Small projects on action recognition or gesture classification.
  • Capstones that start from students’ own recordings instead of static datasets.
  • Metadata labels make it easy to organize and batch-process student-collected data.
Biomechanics & PT / OT

Movement & Gait

  • Comparing joint trajectories between conditions or techniques.
  • Teaching kinematics with simple stick-figures and time series.
  • Linking qualitative observation to basic quantitative measures.
  • Activity and camera-angle tags keep clinical data organized.
Sports Science & Performance

Technique & Consistency

  • Looking at form in jumps, cuts, pivots, or landings.
  • Exploring left–right differences or stability over time.
  • Letting students try simple performance metrics of their own.

From Video to Labeled Data, Briefly

The workflow is straightforward. The goal is to get out of the way so students can focus on the ideas you’re teaching.

  • 1. Capture or import. Film on an iPhone or use an existing clip.
  • 2. Check the visualization. Quick visual QC with sticks, capsules, or sprites. Adjust confidence and smoothing for obvious spikes/noise.
  • 3. Tag metadata. Label the activity, camera angle, and target subject before export.
  • 4. Export for analysis. Use the metadata-enriched JSON in notebooks and the MP4 in reports/presentations.

Python Tools (Optional)

For classes that already use OpenPose-style tooling, we maintain two small scripts that help bridge MovementModeler exports into common workflows. 

# 1) MovementModeler clip export -> per-frame OpenPose-style JSON
python scripts/mm_to_openpose.py movementmodeler_clip.json frames_json/

# 2) Per-frame JSON -> stick-figure MP4
python scripts/mm_json_dir_to_stickmp4.py frames_json/

Repo: movementmodeler-tools

If you’d like to see it in action

A short screen capture showing the basic flow: import a clip, inspect the visualization, tag with metadata, and export MP4 + JSON.

Quick notes for best results

  • Stable framing helps.
  • Higher resolution and less motion blur generally track better.
  • Outdoor clips in bright sunlight work well.
1-minute demo

MovementModeler Demo Thumbnail
Left: in-app workflow · Right: exported stick-figure MP4
Tap to play fullscreen.

Instructor FAQ

Where can students use MovementModeler?

Anywhere they can film. The app works indoors and outdoors, including in direct sunlight. Steady framing and clear subjects help the results.

Do students need special hardware?

Just a reasonably recent iPhone. Analysis can happen in whatever environment you already use — Python, MATLAB, etc.

What does the export look like?

Exports include stick-figure MP4s and BODY-25 style pose data with coordinates and confidence values per joint, per frame, per person. The JSON also includes an lmm metadata block with the activity label, camera angle, and target subject index you set before exporting.

What’s the metadata for?

It makes your data self-describing. Instead of tracking which file is which in a spreadsheet, every export carries its own activity label, camera angle, and subject index. This is especially useful when students collect multiple clips across conditions or sessions.

Can we build our own tools on top?

Yes. The intent is that you and your students treat the exported data as raw material. The optional tools repo exists to make it easier to slot into existing workflows.