Overview

What Upload Tracking does, when to use video vs image, and how to start.

Upload Tracking handles video and image uploads for pose analysis.

For videos, Upload Tracking can work either with a built-in exercise or with a custom reference movement created in the dashboard. For full setup and returned data, see REFERENCE.

Need a faster implementation path?

We also provide a GitHub repo with LLM-ready prompts and examples for uploaded workout video analysis, reference comparison, and AI coaching: https://github.com/Movelytics/posetracker-llm-prompts

Useful for vibe coders building:

• Flutter video analysis

• TypeScript upload flows

• AI coaching review screens

Base URL:

https://app.posetracker.com/pose_tracker/upload_tracking

What you can do

Video (source=video, default)

• Upload a video file, or pass a remote URL (videoSrc).

• Run pose estimation in real time.

• Optional skeleton overlay (skeleton=true).

• Optional movement tracking:

  • built-in exercise tracking with exercise=...

  • custom reference-based tracking with reference=REFERENCE_UUID This can return reps, posture state, progression, recommendations, or reference comparison scores depending on the selected mode.

• Optional export: generate a new video with the overlay.

Image (source=image)

• Upload an image file, or pass a remote URL (imageSrc).

• Run pose estimation once.

• Optional skeleton overlay.

• Optional keypoints + angles output.

• No exercise/movement tracking (no counter, progression, recommendations).

WebView usage

You can embed the endpoint page in a WebView. The page sends results back to the host app via sendDataToNative-style messages.

See WebView / native messages.

Quick start

Video (user uploads a file)

https://app.posetracker.com/pose_tracker/upload_tracking?token=YOUR_API_KEY&source=video&exercise=squat

Image (user uploads a file)

https://app.posetracker.com/pose_tracker/upload_tracking?token=YOUR_API_KEY&source=image

When to use what

• Count reps on a video:

  • source=video

  • upload or videoSrc=...

  • exercise=...

• Compare a video against a custom reference movement:

  • source=video

  • upload or videoSrc=...

  • reference=REFERENCE_UUID

• Export an overlay video:

  • source=video

  • export=true

  • exercise=...

• Analyze a single image (keypoints/angles/overlay):

  • source=image

  • upload or imageSrc=...

  • keypoints=true and angles=true (plan-gated)