How do i label every unlabeled pixel automatically in the Image Labeler App?

I have several pictures from an Infraredcamera, that observes bubbles that occur during a boiling process. The area of interest are only the bubbles. The background/noise doesn't matter. I want to label these pictures and train a deep neural network to identifiy those bubbles. Therefore i labeled images in the Image Labeler App.

Example of unlabeled and labeled pictures:

I only labeled the bubbles and nothing else, so there is only the class "Bubbles". The problem that occurs now, is that the training networks don't accept one class, rather they need more than one. How can i solve this problem? Is there a possibility to automatically label every unlabeled pixel and issign the class "Background" to them?

Answer 1

[Nitin Kapgate]

In a pixel-labelled image, each pixel can have at most one pixel label. When you apply a label to a pixel, the new label replaces the previous label.

Use the following steps to ensure that all pixels in an image are labeled:

1. Begin by labeling the entire image with a single label. Pick a label that represents a predominant ROI in the image, such as background.
2. Then, use the labeling tools to relabel objects with their correct label.

Alternatively, if you have already labelled the desired pixels, to fill all or all remaining pixels, select an ROI label from your list and press Shift+Click (you can use left- or right-click).

Here is an explainer about labeling pixels for Semantic Segmentation using the Image Labeler App.