Local dataset quality check
Validate segmentation masks before training
Segmentation labels can contain malformed polygons, self-intersections, invalid bounds, or masks that do not match their images.
What to check
Inspect polygon and mask findings locally before a segmentation training run starts.
Why it matters
Broken masks teach the wrong boundaries and can cause training failures or misleading pixel-level metrics.
When Pro helps
Pro adds saved repair sessions and bulk editing for repeated cleanup work.
What your report can look like
Segmentation mask review
An illustrative example of the findings and repair guidance Labelyst produces for your dataset.
Status: Blocked
Score: 64/100
2 findings
Findings
- 2 polygons extend outside image bounds
- 1 mask has an invalid image pairing
Passed check
- Valid masks were decoded locally
Suggested next step: Fix mask geometry and image pairing 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.