vars-annotation 0.4.1

This release adds support for annotation localization (e.g. drawing bounding boxes on video) via Cthulhu. Documentation for this is here

Bounding boxes are being stored as an Association with the following structure:

• mime_type: application/json
• link_name: bounding box
• to_concept: self
• link_value:

{
  "x" : 32,
  "y": 120,
  "width": 100,
  "height": 200
}

The x, y, width and height have units of pixels in image coordinates. Image coordinates have an origin (0, 0) at the upper left corner. +x is right, +y is down. The pixels are values from the original un-scaled video.