When using the Interactive Annotator node, the image dimensions are always remembered from the first run for any given row ID, and are not updated after resetting the node and running again.
To reproduce, run the attached workflow, then reconfigure the image dimensions in the Image Generator, run the workflow again, and compare the input image dimensions with the dimensions of the labeling output.
While this is no problem when using unique row IDs derived e.g. from file names, it gets problematic when the annotation input is an image generated by the Create Image aggregation method in a GroupBy node.
(I think there was a similar issue discussing annotated regions being persistent across multiple runs, also depending on the RowID. Sorry if this is a duplicate.)
When using the Interactive Annotator node, the image dimensions are always remembered from the first run for any given row ID, and are not updated after resetting the node and running again.
Interactive Annotator Dimensions.zip
To reproduce, run the attached workflow, then reconfigure the image dimensions in the Image Generator, run the workflow again, and compare the input image dimensions with the dimensions of the labeling output.
While this is no problem when using unique row IDs derived e.g. from file names, it gets problematic when the annotation input is an image generated by the Create Image aggregation method in a GroupBy node.
(I think there was a similar issue discussing annotated regions being persistent across multiple runs, also depending on the RowID. Sorry if this is a duplicate.)