cvat-ai / cvat

Annotate better with CVAT, the industry-leading data engine for machine learning. Used and trusted by teams at any scale, for data of any scale.
https://cvat.ai
MIT License
11.74k stars 2.88k forks source link

YOLO format may export incorrectly if there are skeletons with repeated sublabels names #8090

Open bsekachev opened 2 days ago

bsekachev commented 2 days ago

Actions before raising this issue

Steps to Reproduce

The task has all labels from Human Pose Estimation model (added first) and rectangle label (added second) In exported file it says rectangle has class 138 when only 74 classes are declared in object.names.

Expected Behavior

Export example: task_crowd 1_annotations_2024_06_26_08_36_21_yolo 1.1.zip

There is a task backup just for information, but after importing the backup, labels will be reordered and problem did not reproduced for me. So, reproducable only in initial task:

Nevertheless backup may be useful.

task_crowd_backup_2024_06_26_08_18_15.zip

Possible Solution

No response

Context

No response

Environment

v2.14.3