Local dataset quality check
Check COCO annotations before training
COCO datasets can hide category drift, broken image references, invalid boxes, polygons, or masks.
What to check
Inspect categories, geometry, media coverage, and split evidence in one local report.
Why it matters
A flawed COCO export can train against the wrong classes or omit data without an obvious error.
When Pro helps
Pro makes recurring review and repair workflows practical.
What your report can look like
COCO annotation export
An illustrative example of the findings and repair guidance Labelyst produces for your dataset.
Status: Ready with warnings
Score: 88/100
2 findings
Findings
- 1 image reference has no matching media
- 2 polygon annotations need review
Passed check
- Category IDs map to unique names
Suggested next step: Restore the missing media reference before training.
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.