Problem in sematnic segmentation error: pixel Label IDs must be unique

Hi,

I am trying to classify a set of images using deep learning, following the example provided by https://it.mathworks.com/help/vision/ug/semantic-segmentation-using-deep-learning.html

However, when I run the camvidPixelLabelIDs function, I get the following error:

Error using pixelLabelDatastore>parseInputs

The value of 'pixelLabelID' is invalid. The

pixel label IDs must be unique.

Error in pixelLabelDatastore (line 210)

[location, classes, values,params] = parseInputs(varargin{:});

Any insights on this? Thank you to anyone who will answer

Answer 1

[Chetan]

Hello @VALERIA CROCE,

I understand that you're attempting to classify a set of images using deep learning, following a example. However, you're encountering an error when running the 'camvidPixelLabelIDs' function, stating that pixel label IDs must be unique

The error you're encountering likely stems from having duplicate values in the pixel label IDs you're providing to the 'pixelLabelDatastore' function. This function creates a datastore for pixel label data, used for semantic segmentation tasks in computer vision.

Here's how to resolve this issue:

• Ensure that the pixel label IDs you're providing to the 'pixelLabelDatastore' function are unique.
• You can verify this by examining the output of the 'camvidPixelLabelIDs' function.

Here's an example:

labelIDs = camvidPixelLabelIDs();
disp(unique(labelIDs));

This will display the unique values of 'labelIDs'.

For more details on 'pixelLabelDatastore' and the 'PixelLabelID' argument refer to following Mathworks Documentation

https://www.mathworks.com/help/vision/ref/pixellabeldatastore.html

I hope this helps!