UltralyticsAssistant
commented
1 month ago @tlong123 hi there! 👋 It sounds like you've done everything correctly by running check_dataset, and it's surprising that the labels aren't recognized after the upload to the Ultralytics HUB. This issue might involve a subtle detail in the dataset structure or labeling format that's not caught by the preliminary checks.
A few things to consider:
- Ensure that your dataset's directory structure strictly follows the expected format, with separate folders for images and labels within each data split (e.g.,
train,val). - Verify that label files match the corresponding image files in name, location, and count.
- Double-check for any discrepancies or formatting errors in your label files that could have been overlooked by the dataset checker.
Since the check_dataset tool indicates successful validation but the upload process is still failing, it might also be a good idea to re-zip and re-upload your dataset, ensuring no files are missing or corrupted in the process.
If you continue to face issues, please provide more details about your dataset's structure and any steps you've taken since encountering the error. For further guidance, the Ultralytics HUB Docs at https://docs.ultralytics.com/hub might offer additional insights into dataset preparation and troubleshooting.
Keep us posted! Your feedback helps us improve the HUB for everyone. 🚀
tlong123
sergiuwaxmann
github-actions[bot]
Search before asking
HUB Component
Datasets
Bug
I have created a new dataset based off of an existing one I have used before. After zipping it I ran the check_dataset function against it and recieved a "checks completed correctly" response, however when I upload it to the hub all I get is "Invalid The dataset labels cannot be located."
the output I get from check_dataset is as follows:
Environment
check_dataset ran on m1 mac pro
Minimal Reproducible Example
No response
Additional
No response