Qoriant Thesis / 002

Partially verified

Human Movement Intelligence: Toward a Foundation Model of the Human Body

Video and biomechanical modeling now make useful movement measurement possible outside specialist laboratories. A general foundation model remains a research direction; task-specific systems with known error bounds are the credible product layer today.

Versioned research judgment
Version0.3
Last updated2026-08-15
01

Executive Summary

Video and biomechanical modeling now make useful movement measurement possible outside specialist laboratories. A general foundation model remains a research direction; task-specific systems with known error bounds are the credible product layer today.

02

The Problem

Movement data, biomechanics and expert interpretation remain fragmented across sport, rehabilitation and daily-life settings.

03

Why It Matters

Accessible measurement could expand functional assessment, longitudinal monitoring and research participation.

04

Why Now

Markerless pose estimation, parametric body models and validated smartphone workflows have lowered capture costs.

05

Scientific Foundations

FACTEvidence A

OpenCap estimates kinematics and dynamics from two or more smartphone videos and reported validation against laboratory measurements.

FACTEvidence A

SMPL-X provides a learned parametric representation for expressive 3D human body capture.

06

Key Technical Approaches

  • Markerless multi-view capture
  • Parametric body models
  • Physics-informed biomechanics
  • Longitudinal movement embeddings
07

Major Papers

APaperPrimary
OpenCap: Human movement dynamics from smartphone videos

Peer-reviewed validation of smartphone-video biomechanics.

PLOS Computational BiologyOpen source ↗
APaperPrimary
Expressive Body Capture: 3D Hands, Face, and Body from a Single Image

Primary paper for SMPL-X body representation.

CVPROpen source ↗
08

Important Researchers

09

Leading Labs

10

Companies

  • Vicon
  • SWORD Health
  • WHOOP
11

Clinical / Commercial Evidence

FACTEvidence A

OpenCap reports field and laboratory validation; this does not make every downstream diagnosis or coaching claim valid.

12

What Has Actually Been Demonstrated

FACTEvidence A

Smartphone video can support task-specific biomechanics estimates under a defined capture workflow.

13

What Has Not Been Demonstrated

UNCERTAINEvidence B

A single representation validated across sport, rehabilitation, aging and robotics has not been established.

14

Technical Bottlenecks

  • Occlusion and camera variation
  • Biomechanical ground truth
  • Cross-body and cross-task transfer
  • Safety-critical interpretation
15

Data Bottlenecks

  • Limited diverse motion datasets
  • Inconsistent labels and protocols
  • Few longitudinal outcome-linked datasets
16

Regulatory Questions

  • When does movement scoring become a medical claim?
  • How should video and biometric data be retained?
17

Venture Landscape

  • Capture platforms
  • Remote rehabilitation
  • Athlete monitoring
  • Movement datasets
18

Potential Venture Directions

  • Video-first functional assessment
  • Outcome-linked movement datasets
  • Human-reviewed coaching infrastructure
19

Qoriant View

QORIANT VIEWEvidence B

The first valuable layer is reliable measurement with scoped claims, not a universal body model.

20

What Can Now Be Built?

  • Low-cost movement assessment for defined tasks
  • A benchmark connecting pose error to biomechanical decision error
  • Consent-aware longitudinal movement data infrastructure
21

Sources

Read the Qoriant Research Standard ↗

APaperPrimary
OpenCap: Human movement dynamics from smartphone videos

Peer-reviewed validation of smartphone-video biomechanics.

PLOS Computational BiologyOpen source ↗
APaperPrimary
Expressive Body Capture: 3D Hands, Face, and Body from a Single Image

Primary paper for SMPL-X body representation.

CVPROpen source ↗