Semantic segmentation evaluation results

Hi

When i try evaluating semantic segmentation results by running "metrics = evaluateSemanticSegmentation(pxdsResults,pximds)" i get this error "size of the categorical data returned by dsResults and dsTruth must be the same". Please how can i fix this

Answer 1

[Vidip]

The error message you're encountering, "size of the categorical data returned by dsResults and dsTruth must be the same," suggests that there's a mismatch in dimensions between the predicted segmentation results (pxdsResults) and the ground truth data (pximds).

To fix it, you need to ensure that all images in pxdsResults and pximds have the same dimensions. If the images or their corresponding segmentation masks have been resized or cropped during preprocessing or prediction, you'll need to make sure that both the predictions and the ground truth data are brought to the same dimensions before evaluation. Also, make sure that the categories (classes) in pxdsResults and pximds match exactly.