About PosturaScreen

We turn two photos into a 17-metric posture screening report — for movement professionals and the people they help.

What we do

PosturaScreen is a browser-based AI posture screening tool. A user takes a front and side photo with any phone (no app install), and our biomechanical AI model detects 17 anatomical landmarks. From those landmarks we compute 17 quantified measurements — head tilt, shoulder level, forward head, pelvic alignment, knee angles, and more — and produce a clean, single-page report your client can read or print.

It's a screening tool, not a diagnostic device. We focus on quantification, repeatability, and clear visualization so that posture findings become objective rather than impressionistic.

Who it's for

The platform was built first for movement professionals — physiotherapists, chiropractors, sports rehab clinicians, personal trainers, and wellness clinics — who need a defensible, repeatable way to communicate posture findings to clients. As the tool matured, we noticed an equally strong use case for individuals who want to check their own posture for awareness, especially desk workers, lifters, and parents curious about their kids' posture as screen time climbs.

Our approach

The goal is straightforward: take something practitioners already do visually, and add a layer of quantified, photographable measurement that they can hand to a client or compare across sessions.

We chose 2-D photo-based screening over sensor wearables or 3-D scanning because the bar to entry has to be near-zero — any phone, any clinic, any browser. The trade-off is honest: a single 2-D photo cannot replace a radiographic measurement. Where the model produces a screening estimate (spine curves, Q angle, pelvic tilt), we tag it with an explicit approx marker on every report.

Technology

The detection layer uses YOLO26-pose to identify 17 anatomical keypoints. The metric layer is purely geometric: we compute ratios, angles, and horizontal offsets between landmarks using formulas grounded in established posture-assessment literature. Outputs flow into a static report with skeleton overlays drawn directly on the user's own photos.

Photos are encrypted in transit and at rest, served only via time-limited URLs, and deletable by the user at any time. We do not train models on user photos without explicit, separate consent.

For the technical detail of which keypoint feeds which metric, see our methodology page.

Team

The PosturaScreen team is a small group of engineers and movement-science enthusiasts. We deliberately keep individual identities private and let the product, the methodology, and the editorial standards speak instead. See our editorial team statement for the reasoning.

Get in touch

Questions, feedback, press inquiries, or partnership ideas: support@posturascreen.com.

For information about how we write and review content, see our editorial standards. For the underlying measurement approach, see methodology.