Skip to content

Home

Jump to bottom
Juliette Regimbal edited this page Apr 8, 2020 · 2 revisions

This wiki is specifically about the Pixel.js job in Rodan. For general information on how to use Pixel.js, consult the main Pixel.js wiki.

Differences from Standalone Pixel.js

The main points of difference between Pixel.js in Rodan and outside of it are:

  1. The number of input and output layers must be defined before starting;
  2. A background layer and selected region(s) layer are exported by default in addition to user defined layers; and
  3. There is a choice of if the exported layers will contain the contents of the source image (default, for patchwise training/classification) or be a mask (for pixelwise training/classification).

For more information on how layers are computed, consult the technical info page.

Use in the Workflow Editor

From the workflow editor, define two more output ports than layers you will classify as two will be used for the background and selected regions layers. Right-clicking on the job box and going into the "Settings" menu will provide a Boolean option "Export Mask". By default this is false and will export layers where the opaque parts will contain the contents of the source image instead of a solid color. Setting it to true will result in a mask being exported.

In the Interactive Phase

To finish the Pixel.js Rodan job, there is a new button "Submit to Rodan". This will signal the end of the job, computing the background layer (that is, the selected region(s) minus the user-defined layers), and sending all layers back to Rodan.

Unlike in earlier versions of the Pixel.js Rodan job, there is no progress bar displayed. The export process only takes a few seconds and upon completion an alert will be displayed informing you that the data has been submitted and Pixel.js can be closed.

Clone this wiki locally