The honest answer is that nobody has properly measured this
If a provider claims verified plus-size accuracy, ask for the study. There is no published subgroup accuracy benchmark for AI body measurement stratified by size, which means claims in either direction are unproven.
What is documented
Several convergent concerns, none of them a quantified bias finding.
A leading 3D scanning provider told Apparel Resources in 2019 that very large bodies make it difficult to achieve clean arm and leg separation, which is needed for good measurement extraction. That is a vendor documenting a limitation of their own technology.
Manual measurement error grows with waist size. Researchers note waist circumference reliability is lower among people with larger waists, and that earlier studies report low reliability of manual waist measurement in obese subjects.
The size chart problem is measurable, though
This part is quantified. Research measuring over 1,000 garments across 50 UK brands found chart accuracy fell with size: 71% at UK 8-10, 68% at UK 12-14, 61% at UK 16-18, and 54% at UK 20-22.
So at the largest sizes measured, nearly half of garments did not match their own brand's published chart within 1 cm.
Shoppers report it too
In a 2025 survey, sizing inconsistencies were cited by 46% of plus-size respondents, against 34% of mid-size and 25% of straight-size shoppers. That is a real difference in experience, measured directly.
Where the datasets come from
Most published validation work samples young, lean Western participants. Researchers explicitly warn that obese, very short or very tall participants affect measurement variability. One large recent study did find waist circumference rising in women at flat BMI, which is itself relevant.
The pattern is consistent across evidence types: larger bodies are described less accurately by charts and are absent from validation sets. That is a reason for caution, not a quantified accuracy figure.
Practical advice
- Treat any claimed plus-size accuracy as unproven until a study says otherwise.
- Weight measurement error grows with waist size, so measure your waist especially carefully.
- Expect the brand's chart to be least reliable exactly where you need it most.
- Rely on your own recorded measurements rather than a one-time inferred size.
The real problem is upstream
For plus-size shoppers the binding constraint is not measurement technology. It is that size charts are least accurate at the sizes where accuracy matters most. AI measurement cannot fix a chart that is wrong.