Due to a bug in the MergeQueue handling during agglo.Rag.learn_agglomerate(), the progress bar has to be turned off. This is done by the statement, g.show_progress = False. This makes it difficult to predict when the learning process will finish.
The problem has to do with miscounting of remaining valid items. My theory is that items are merged that are not in the merge queue. Therefore the number of nodes decreases but not the number of valid items.
Due to a bug in the MergeQueue handling during agglo.Rag.learn_agglomerate(), the progress bar has to be turned off. This is done by the statement,
g.show_progress = False. This makes it difficult to predict when the learning process will finish.The problem has to do with miscounting of remaining valid items. My theory is that items are merged that are not in the merge queue. Therefore the number of nodes decreases but not the number of valid items.