Local dataset quality check
Audit a dataset from an annotation vendor
A vendor delivery can meet a file-format contract while still containing label drift, incomplete coverage, or geometry errors.
What to check
Run a local acceptance check for class coverage, media integrity, image quality, split evidence, and annotation geometry.
Why it matters
Catching quality gaps before accepting a delivery prevents expensive rework after training has already begun.
When Pro helps
Pro makes repeated acceptance reviews and repair sessions practical.
What your report can look like
Vendor delivery acceptance review
An illustrative example of the findings and repair guidance Labelyst produces for your dataset.
Status: Ready with warnings
Score: 83/100
2 findings
Findings
- 2 annotations need acceptance review
- 1 class has partial split coverage
Passed check
- Delivered media inventory was readable
Suggested next step: Return the documented repair list to the vendor.
Rules: standard profile · local check
Illustrative local result. It is not a certification.
Check it before training
No account or upload is required for a Free check. Analysis, repairs, and reports stay on your device.