DecodeTransform

Decodes images


DownloadTransform

Downloads an image from a URL


GalleryOutputTransform

DOCUMENT ME CHARLES


IncrementalOutputTransform

Incrementally output templates received to a gallery, based on the current filename

When a template is received in projectUpdate for the first time since a finalize, open a new gallery based on the template's filename, and the galleryFormat property.

Template received in projectUpdate will be output to the gallery with a filename combining their original filename and their FrameNumber property, with the file extension specified by the fileFormat property.


OpenTransform

Applies Format to Template filename and appends results.


OutputTransform

DOCUMENT ME CHARLES


PrintTransform

Prints the file of the input Template to stdout or stderr.


ReadLandmarksTransform

Read landmarks from a file and associate them with the correct Template.

  • file: io/readlandmarks.cpp
  • inherits: UntrainableMetadataTransform
  • author(s): Scott Klum
  • format: Example of the format:
    image_001.jpg:146.000000,190.000000,227.000000,186.000000,202.000000,256.000000
    image_002.jpg:75.000000,235.000000,140.000000,225.000000,91.000000,300.000000
    image_003.jpg:158.000000,186.000000,246.000000,188.000000,208.000000,233.000000
    
  • properties: None

ReadTransform

Read images


WriteTransform

Write all mats to disk as images.

  • file: io/write.cpp
  • inherits: TimeVaryingTransform
  • author(s): Brendan Klare
  • properties:

    Property Type Description
    outputDirectory QString Top level directory to write images to.
    underscore QString
    imgExtension QString Extension to save images as.
    padding int Zero pad the image filenames.
    subDir QString Optional group images by a metadata key in subdirectories.

YouTubeFacesDBTransform

Implements the YouTubesFaceDB

  • file: io/youtubefacesdb.cpp
  • inherits: UntrainableMetaTransform
  • author(s): Josh Klontz
  • read:

    1. Wolf, Lior, Tal Hassner, and Itay Maoz.
      "Face recognition in unconstrained videos with matched background similarity." Computer Vision and Pattern Recognition (CVPR), 2011 IEEE Conference on. IEEE, 2011.
  • properties: None