Training semantic segmentation on grayscale image gives RGB-Triplet pixel label IDs error

I am following the example under "Train a Semantic Segmentation Network" at https://www.mathworks.com/help/vision/ug/semantic-segmentation-examples.html#mw_c2dc3c5f-86f8-4835-be60-9799bfe66ced

However, I am trying to run the code on my set of images with are grayscale instead of rgb

I just tried

imageDir = fullfile(outputDir,'myImages');
labelDir = fullfile(outputDir,'myLabels');
imds = imageDatastore(imageDir);
classNames = ["objects","background"];
labelIDs   = [255 0];
pxds = pixelLabelDatastore(labelDir,classNames,labelIDs);
I = read(imds);
C = read(pxds);

But that gives me this error:

Error using matlab.io.datastore.PixelLabelDatastore.label2cat (line
970)
Pixel label image has RGB-triplet pixel label IDs instead of scalar
pixel label IDs.
Error in matlab.io.datastore.PixelLabelDatastore/label2categorical
(line 878)
                        L =
                        matlab.io.datastore.PixelLabelDatastore.label2cat(...
Error in matlab.io.datastore.PixelLabelDatastore/read (line 459)
                C = this.label2categorical(C, info);
Error in 
C = read(pxds);

Why is this?

Answer 1

[Alexandre Broc]

Hi, did you find an answer ?

I got the same problem

[Raza Ali]

Have you got the answer? or got the solution to this problem?

[yuwen liu]

Hi, did you find an answer ?

I got the same problem.

My label images were RGB.But when I ran "[net, info] = trainNetwork(pximds,lgraph,options); ",

I had the error

wrong use trainNetwork (line 165)

Pixel label image has RGB-triplet pixel label

IDs instead of scalar pixel label IDs.

[Curtis Zhang]

delete alpha and transfom 3d into 2 d