±2%
±2% Accuracy Range
Validated against clinical Dual-Energy X-ray Absorptiometry (DEXA) scans.
DEXA Validation
DEXA-Grade Body Composition Accuracy via Computer Vision
Discover how PoseScore achieves a ±2% margin of error against clinical DEXA scans, without green screens, specialized hardware, or clinic visits.
Proof at a Glance
Key Validation Metrics
0
Additional Hardware
Runs entirely on standard camera imagery without depth sensors or green screens.
DL
Deep Learning Precision
Advanced neural networks trained directly on DEXA ground-truth datasets.
White Paper Highlights
Core Pillars of DEXA-Aligned CV
Gold-Standard DEXA Calibration
While BMI fails to distinguish muscle from fat and consumer BIA scales fluctuate with hydration, PoseScore's models are trained directly on DEXA scan baselines. Our neural networks map morphological fat distribution with clinical-level precision.
Evolving Beyond Legacy Research
Early academic studies (such as the 2018 PLOS ONE study) proved 2D image body fat estimation was possible, but required green screens, rigid 91-inch measurements, and green-wrapped limbs. PoseScore elevates this concept into a seamless commercial tool-delivering higher precision without environmental constraints.
Unmatched Commercial Precision
Across diverse demographics and body types, our vision algorithms maintain a tight ±2% correlation bracket with DEXA results. Whether monitoring athletic performance or general wellness, PoseScore provides trustworthy metrics at scale.
Workflow: Simple Photo-Based Calculation
How It Works: The User Workflow
A simple 3-step experience that transforms 2D photos into DEXA-grade body composition insights.
Advanced Body Fat Calculator
Step 01
Upload 3 Standard Photos & Metrics
Provide front, side, and back profile photos alongside basic parameters (Height, Weight, DOB). No green screens, wrapped limbs, or rigid camera distances required.
Step 02
Computer Vision Inference
When “Calculate BodyFat” is selected, PoseScore’s neural network performs semantic segmentation, isolating contours and mapping fat distribution patterns.
Step 03
DEXA-Calibrated Output
Receive an instant body fat percentage estimate, rigorously validated to fall within a ±2% margin of error against a true clinical DEXA scan.
Validation Data
Data Preview
Correlation Scatter Plot: PoseScore CV Model vs. True DEXA Scan
Tight clustering along the trendline proves high statistical correlation across all body fat percentages.
Error Distribution Plot
Over 90% of model predictions fall strictly within the ±2% variance threshold.
Read the Full Technical White Paper
Get an in-depth breakdown of our validation methodology, dataset composition, and comparative benchmarks.
Download White Paper (PDF)