Video Annotation Has Never Been Easier

vatic.js uses optical flow to help automate video annotations. Get started below!

: Upload Dataset

Select a zip archive of frame images.

The zipped archive should not contain a folder, only the jpg images.

Images should be named with no leading zeros: <frame num>.jpg

You can create the archive with: $ zip frames.zip *


If you would like timestamping (for time syncing) make sure to add a timestamp.xml in the frame zip archive.


Optionally, you can load an existing vatic.js XML annotation file to pick up where you left off or to correct a previous annotation.

: Annotate Frames

Annotate a frame sequence above.

Press Alt+n to create a new bounding box. Then click two positions on the video corresponding to the top-left and bottom-right corners of the box.

Tips:

  • Use the spacebar to play/pause the video and the and keys to move frame-by-frame.
  • If an object disappears from the image, toggle the visibility checkbox.
  • Zoom in with your browser to place the bounding boxes more accurately.
  • While seeking, holding down ctrl will mark all objects as not visible and prevent optical flow tracking (makes seeking faster).