Data annotation format

Please provide the following information when requesting support.

• Hardware (RGX 3080)
• Network Type (Classification)
• TLT Version format_version: 2.0
toolkit_version: 3.22.05
published_date: 05/25/2022()
• Training spec file
• How to reproduce the issue ?

Hi there.

I am experiencing an UnknownImageError and I would like to check that my naming convention is not the cause:

Does this mean that I need to put “–” in front of each .jpg name?
Does it mean that I need to put “–” in front of each class name folder?
Does it mean that I need to put “__” in front of “train”, “val” and “test” directory names?
I note that the datasets in the tao-voc exercise were not named this way.
My 2 classes are currently named “1.jpg” - “502.jpg” and “01.jpg” - “0502.jpg”.
Have I got this wrong?

No, it’s just an indicator of folder relation.
You can check how the files are splitted into train/val/test in “tao-voc/classification.ipynb”, it just uses the normal folder name and file name.

Okay, thank you.

And my naming convention is okay?

There is no update from you for a period, assuming this is not an issue anymore. Hence we are closing this topic. If need further support, please open a new one.

I think the naming is fine, basically the script uses glob.glob to get the “*.jpg” file names in the source folder as a filename list and then shuffle and split to different folders, there is no specific naming rule for the file names.

This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.