Annotation classes incremented when exporting dataset

Uploading downloaded dataset folder to new project which has Json coco annotations with it.

Dataset images upon uploading with annotations gives issue.

For e.g: my dataset classes are “0,1,2,3,4,5,6,7,8,9,.,kwh”.

when checking the bounding boxes fore.g:
bounding box over 0 gives 1 class annotation, similarly for 1 gives 2 and for 2 gives 3.

So the issue is class is one step incremented how to fix that.

Screenshot 2024-04-04 083337

see in this image 1 is shown over annotation of 0 digit.

I solved the issue, the problem was after extracting zip I get 3 folders train,test,valid and 2 extra txt files i.e readme.txt and requirements.txt.
when uploading whole folder after extracting to my project extra classes were added by txt files.
I removed them and the classes were accurate but still incremented. i.e on 0 digit like above picture showed 1 for some reason.

I open annotation coco json files of train,test and valid and start changing with classes by moving it one place or adding one new class. After trail and error i fixed it. I had to manually update and check the annotation class to finally see the results.

This topic was automatically closed 7 days after the last reply. New replies are no longer allowed.