NL | EN

Database

Strategic

The database that didn't exist yet

Babon is building a large, labelled clinical motion database. The figures below come live from the research database.

Live Recordings in database 71.400 Aggregated from public, CC-BY licensed datasets. Base layer for normative comparison and validation.
Unique subjects 549 From public, CC-BY licensed sources. From practices and clinics: not yet.
Data sources 3 Public datasets, all CC-BY licensed and fully attributed.
Infrastructure fr-par Scaleway Paris. Data does not leave the EU.

Updated: 23 April 2026 · source: research.movalytics-db

Composition of the 71,400 recordings

// breakdown per bron
AddBiomechanics · 69.804 GAVD · 1.462 LabValidation · 144
AddBiomechanics 97,8%
GAVD 2%
LabValidation 0,2%
Why this database

Context

A large, representative clinical motion database from everyday practice does not currently exist. The only conventional source is a gait lab, where a single measurement costs tens of euros. As a result, analyses like age-norms, recovery patterns after knee surgery, and diagnosis-specific reference data are simply not feasible in practice.

Babon starts with three publicly available, CC-BY licensed datasets, fully attributed below. Because a Movalytics analysis costs cents per recording, the database can be gradually expanded over time through anonymised contributions, with no footage and no patient data.

Source data

With thanks to

Every recording in the research database comes from publicly available, CC-BY licensed datasets. Full attribution below.

[01] Normative reference · healthy population

AddBiomechanics

Werling, K., Bianco, N.A., Raitor, M., Stingel, J., Hicks, J.L., Delp, S.L., Liu, C.K. (2023). AddBiomechanics: Automating model scaling, inverse kinematics, and inverse dynamics from human motion data through sequential optimization. PLoS ONE 18(11): e0295152.

Aggregated marker-based motion capture across 15 contributing studies. Base layer for normative comparison.

+ Contributing studies 15
  1. Lencioni et al. 2019 50 subjects
  2. Carter et al. 2023 50 subjects
  3. Santos et al. 2017 49 subjects
  4. Camargo et al. 2021 22 subjects
  5. Tan et al. 2023 17 subjects
  6. Moore et al. 2015 12 subjects
  7. Falisse et al. 2016 11 subjects
  8. Van der Zee et al. 2022 10 subjects
  9. Hamner et al. 2013 10 subjects
  10. Uhlrich et al. 2023 10 subjects
  11. Tan et al. 2022 9 subjects
  12. Wang et al. 2023 9 subjects
  13. Han et al. 2023 7 subjects
  14. Fregly et al. 2012 6 subjects
  15. Li et al. 2021 1 subjects
273subjects
69.804lab-mocap trials
CC BY 4.0 addbiomechanics.org →
[02] Clinical and atypical gait

GAVD, Gait Abnormality Video Dataset

Ranjan, R., Ahmedt-Aristizabal, D., Ali Armin, M., Kim, J. (2025). Computer Vision for Clinical Gait Analysis: A Gait Abnormality Video Dataset. IEEE Access 13: 45321, 45339. doi:10.1109/ACCESS.2025.3545787

Video-based gait recordings with clinical labels. Supports evaluation of pose extraction on in-the-wild video, beyond lab conditions.

276subjects
1.462videos
CC BY 4.0 IEEE Access →
[03] Video ↔ mocap pairs · validation

LabValidation (OpenCap, Stanford)

Uhlrich, S.D., Falisse, A., Kidziński, Ł., Muccini, J., Ko, M., Chaudhari, A.S., Hicks, J.L., Delp, S.L. (2023). OpenCap: Human movement dynamics from smartphone videos. PLoS Computational Biology 19(10): e1011462.

Calibrated video ↔ mocap pairs as gold standard when comparing Movalytics output against classical gait-lab measurements. Subjects are a subset of AddBiomechanics.

144video pairs
subjects shared with AddBiomechanics
CC BY 4.0 opencap.ai →
Research partnership In conversation with Hogeschool Utrecht, Jaap Jansen, Institute of Movement Studies. The setup is being explored jointly; no data has been contributed yet.

Own database or research question?

Get in touch