Skip to content

Latest commit

 

History

History
27 lines (17 loc) · 1.55 KB

SUPPORT.md

File metadata and controls

27 lines (17 loc) · 1.55 KB
Raw

Support

Welcome to the support page for the Pose Data Annotator app. You can find the most frequent questions here.

If you have a question, feel free to contact us at matyas.bohacek@matsworld.io and thanh.dolong@gmail.com. If you have a suggestion, feature request, or want to file a bug report, proceed here.

How to annotate a single video?

  1. Open the app.
  2. Select the Annotate video section in the left sidebar.
  3. Drop a file into the box, or select one using the Load Video button.
  4. Tap Start Processing and wait for the results. A popup will open once the process is done and you will be able to save the resulting csv file.

How to annotate a dataset?

  1. Open the app.
  2. Select the Annotate dataset section in the left sidebar.
  3. Drop a folder into the box, or select one using the Load Dataset button.
  4. Tap Start Processing and wait for the results. A popup will open once the process is done and you will be able to save the resulting csv file.

How does the app work?

The app utilizes the Vision engine for body, hand and face pose estimation. The videos are split into frames individually annotated and outputted for any potential experiments or use-cases involving pose data.

How to customize the annotations?

Open App Preferences (by pressing Cmd + ,). Here you can customize the FPS to be used, which parts of the body to annotate, and more.