ai engine
AI Engine
Ensemble models estimate body landmarks and volumetric shape from multi-view imagery. Outputs include uncertainty bands—not single false-precision numbers.
Technology
Our stack balances on-device guidance with cloud reconstruction—always with uncertainty bands, never false precision.
ai engine
Ensemble models estimate body landmarks and volumetric shape from multi-view imagery. Outputs include uncertainty bands—not single false-precision numbers.
computer vision
Pose estimation and silhouette extraction stabilize framing so weekly scans remain comparable.
body mapping
A parametric mesh is fit to observed contours, then refined for circumference extraction along anatomical planes.
privacy
TLS in transit, encryption at rest, role-based access for family and trainer shares. We do not sell personal scan data.
accuracy
In controlled tests, key circumferences typically fall within roughly ±1–2 cm of tape measure baselines for average adult builds. Results vary with clothing, lighting, and pose. Forma is not a medical diagnostic device.
ml
Models are validated on diverse body shapes. We continuously evaluate bias and publish methodology notes for enterprise customers.
cloud
Heavy reconstruction jobs run in regional cloud workers for speed; you can request deletion of cloud copies anytime.
edge
Pose guidance and preview run on-device so you get real-time feedback before upload.
encryption
AES-256 at rest for stored artifacts; modern TLS for APIs. Access logs available for business accounts.