Ensuring that Extended Reality (XR) environments are age-appropriate is an important regulatory and safety challenge. However, current age assurance operates only at registration and cannot verify the age of the active user during a session. Face-based approaches, the dominant solution in social media and adult platforms, are impractical in XR, because they require removing the headset and taking a self-captured image, often on a mobile app. This both breaks immersion and introduces the privacy risk of sharing face pictures with third parties, which leaves XR platforms without a viable path to continuous, in-session and privacy-preserving age assurance. We propose the dorsal part of the hand as an alternative to the face, by exploiting the egocentric cameras that XR headsets inherently and naturally use to capture gesture interactions. To evaluate this, we collect an age- and sex-stratified, ethnodiverse dataset of 436 participants spanning the minor--adult boundary, captured under unconstrained lighting and orientation conditions. To characterise what is achievable with off-the-shelf methods at the minor--adult boundary, we evaluate standard neural network architectures for age assurance at the legally critical 18-year threshold. Analysis confirms performance is robust to skin-tone variation. On this dataset, the challenge-31 operating point achieves zero minor admission, making the system a viable first-stage filter for age assurance. These findings position dorsal hand morphometrics as an effective and more privacy-preserving biometric modality for in-session age assurance in XR.
Figures & tables
Figure 1: Two age-assurance modalities in XR. (a) Face-based verification requires the user to lift the headset and present their face to a mobile camera: an out-of-band step that breaks immersion. (b) Dorsal hand inference uses the headset’s own pass-through RGB sensors to capture the hand during normal interaction, requiring no interruption to the XR session.
Figure 2: Overview of the image capturing procedure. Each session comprised a protocol of 16 captures per participant, systematically varying users position relative to light source and two intensity levels to yield eight distinct lighting conditions, with each hand photographed once per condition. This photometric diversity was designed to yield a dataset ecologically representative of consumer-grade capture conditions, contrasting with the controlled clinical photography typical of dermatological studies [ 13 , 30 ] .
Figure 3: Capture frustum and depth-shader visual feedback used to guide hand positioning during data collection. (a) The capture frustum defines the 3D volume within which the hand must be placed for a valid capture. Dimension a denotes the target distance from the camera (30 cm) and b the acceptable depth tolerance ( ± 10 cm). (b) Hand positioned outside the lateral boundaries of the frustum: the depth shader is dark, indicating no valid capture. (c) Hand within the frustum laterally but too far from the sensor: the shader renders partially dark, prompting the participant to move closer. (d) Hand correctly placed at the target depth within the frustum: the shader renders fully white, confirming a valid capture is in progress.
Figure 4: ( a ) Age-band composition of the patch cohort by merged skin-tone group (light = ITA I–III, medium = IV, dark = V–VII; n=428 ). Individual ages are shown from 10–30 years, with grouped bands for ages 31+. ( b ) Age-band distribution of the same participants by sex ( n=428 ). ( c ) Per-participant CIE L∗ – b∗ skin-colour scatterplot for the patch cohort, with points coloured by measured patch RGB. Dotted lines mark the ITA boundaries between the merged light, medium, and dark groups, and the in-panel labels report the proportion of participants in each group.
Figure 5: Age regression results across architectures. Top: MAE by decade. Bottom: Overall MAE, RMSE, and intra-user STD (mean ± std across 5 held-out test folds).
Model
MAE (yr)
RMSE (yr)
Intr-usr STD
SwinV2-B
5.78 ± 0.64
7.47 ± 0.72
2.89 ± 0.20
EffNetV2-S
6.01 ± 0.11
7.97 ± 0.13
3.28 ± 0.19
RN50
6.07 ± 0.43
7.91 ± 0.46
3.51 ± 0.30
MobNetV3
6.96 ± 0.79
9.02 ± 0.87
3.87 ± 0.24
Table 1: Age regression performance (mean ± std, 5 folds).
Figure 6: (a) ROC curves for the binary adult/minor age gate across architectures (mean ± std, 5 held-out test folds). The dotted line marks FPR=0.1 . (b) SwinV2-B ROC curves disaggregated by skin-tone group (Light = ITA I–III; Medium = IV; Dark = V–VII). Shaded bands show std across 5 folds.
Table 2: Age-gate results. (A) Cross-model AUC and [email protected] (mean ± std, 5 folds). (B) SwinV2-B FPR (minors) and FNR (adults) at three thresholds (mean, 5 folds).
τ
Φ−1(1−τ)
Teff
NIST analogue
0.775
−0.76
23.2
Challenge-23
0.923
−1.43
27.9
Challenge-28
0.960
−1.75
30.1
Challenge-31
Table 3: Approximate NIST Challenge- T equivalents for each decision threshold τ , derived via Eq. 4 with σˉ≈7.2 years (inferred from SwinV2-B MAE = 5.78 yr under the Gaussian calibration assumption; see text).
(A) Age-gate ROC by skin-tone group (mean ± std, 5 folds)
Vasco Xu and Henry Hoffmann are with the University of Chicago. · Brian Chen is with Northwestern University. · The remaining co-authors are with Google. +1
1Instituto de Computac¸˜ao, Universidade Estadual de Campinas (UNICAMP), Campinas, Brazil · 2Instituto de Ciˆencias Matem´aticas e de Computac¸˜ao, Universidade de S˜ao Paulo (USP), S˜ao Carlos, Brazil