Workflow

[aisaiahpellecer]

Work Flow

Step 1: API Bulk Work, Generate Images and Metdata JSON

• Saving the API response from this step that will be used Step_3 for the manifest.
• Pulling in the metadata

Step 2: Cutout, Resource Object Detection and Segmentation for UI

Combination of the object detection and segmentation. Would output the masks and the 'cutouts'

• Original image as a thumbnail -- (480 on the longest side as jpegs)
• Output the individual cutouts as png (include, label in filename: image_{resourceid}{objectname}{instance__#} )
  • Normal the sizes of the individual cutouts (max_size= 480, on the longest side )
• Thumbnails of the binary masks: (include, label in filename: mask_{resourceid}{objectname}{instance__#} )
• Output the bounding box information -- JSON output (include, label in filename: image_{resourceid}{objectname}{instance__#}.JSON )
  • Including the four coordinates (Normalized 0-1)

Note: the outputs will be directed into the UI folders.

Step 3: Generate Manifest

Combining the JSON Metadata into a manifest of all of the different assets that were generated.

• In the manifest, each row is an image/ resource. Including of all the metadata below (+ resource_id, item url) and a list of segments.

• Each row of the manifest should include:

  • Item Title
  • URL
  • subject List
  • date
  • alt_text entry -- duplicate the item title for now.
  • contributor_names
  • Locations
  • Collection Name (source_collection)
    • set -- specific to the Free-to-Reuse
  • original_format
  • thumbnail (filename)
  • List of Segements (Filename of the cutout, Filename of the mask, and bounding box information)
    • outputs from step 1 and step 2 (filename and bounding box information)

• Starting with a dummy output