Skip to content

Workflows

Standard Video-First Workflow

Annolid supports direct video labeling without mandatory frame extraction.

  1. Prepare your video data.
  2. Open video in Annolid.
  3. Jump to a representative frame.
  4. Label instances on that frame.
  5. Run tracking/propagation.
  6. Review and correct predictions.
  7. Export annotations or downstream training data.

Practical Notes

  • Use extraction tools only when you explicitly need frame-level datasets.
  • Keep annotations incremental and review high-motion sections early.
  • Prefer short review loops over long unattended tracking runs.

Advanced References