Is there a preferred way to delete a class from a dataset?
I wrote a quick python script to change all the *.txt file class numbers, as well as obj.name, but this seems to often cause conflicts with the JSON's that Darkmark uses. I've tried deleting all the JSON's and regenerating, but that caused even more mislabeling than before. There has to be an easier way, what is your workflow like to achieve this?
I would have said delete the json files and re-import. That is what I do. If you say this causes more problems, then you did something wrong, like possibly forget to update the .names file?
Is there a preferred way to delete a class from a dataset?
I wrote a quick python script to change all the *.txt file class numbers, as well as obj.name, but this seems to often cause conflicts with the JSON's that Darkmark uses. I've tried deleting all the JSON's and regenerating, but that caused even more mislabeling than before. There has to be an easier way, what is your workflow like to achieve this?