Import classes from CSV
complete
This post was marked as
complete
The feature was released this morning 🎉. You can give us your feedback on the feature in the comments!
Nicolas Draber
in progress
If you have a list of hundreds of classes, you will soon be able to add them to one of your dataset by a simple copy-paste
Nicolas Draber
Merged in a post:
Class Creator Tool
D
Dexter Lewis
I have a large list of classes and sub-classes. The current workflow requires I add them manually. Also, it'd be nice to add subclasses to maintain a hierarchical relationship.
Nicolas Draber
Hi Dexter, thanks for your post. I'll merge it with this one: https://labelflow.canny.io/feature-requests/p/import-classes-from-csv
Regarding the subclasses, you can put a "/" between classes and subclasses: "Pole/Wood/Cracks" for instance. Would this meet your needs?
D
Dexter Lewis
Nicolas Draber:
Nick, The "/" would work. The example schema you recommended is "object/material/defect."
Ex. Pole/Wood/Cracks or Insulator/Porcelain/Cracks or Crossarm/Wood/Cracks.
However, there are other data I'd like to label. "object/type", "object/material", "object/condition", "object/configuration." Without a parent-child hierarchal relationship in the editor, would I have to create four labels of the same object to capture all this information? Does that result in 4 bounding boxes or polygons?
Or
Would I have to create a unique combination list? My fear is that could be too large to easily navigate.
object/type1/material1/condition1
object/type1/material1/condition2
object/type1/material2/condition1
object/type1/material2/condition2
....
D
Dexter Lewis
Object/Conditions: Arrester/Healthy + Arrester/Broken + Arrester/Cracked _+ Arrester/Burnt + Arrester/Missing_Animal_Guard + Arrester/Isolator_Blown + Arrester/Misaligned + Arrester/Missing
Object/Type: Arrester/Distribution + Arrester/Riser Pole
Object/Material: Arrester/Porcelain + Arrester/Polymer
D
Dexter Lewis
Last comment...These labels get even more complicated when you begin to add in information related to truncation or occlusion.