Non-Linear Changes in Face Availability During Naturalistic Playtime Across the First Years: Insights From Head-Mounted Cameras and Automated Face Detection.

Faces provide crucial input for early development. This study leveraged innovations in wearable head-mounted cameras (headcams; specifically, TinyExplorer gear) and automated face detection (RetinaFace) to characterise the everyday visual availability of faces during playtime in the home environment across the first years. Using a cross-sectional developmental trajectory design, we collected egocentric headcam data from 29 young children across the first 3 years of life (2-30 months). The dataset comprised 1,891 minutes of video (over 5.5 million frames). We examined cross-sectional developmental trajectories in face availability, spatial distribution, size and size variability. We observed distinct non-linear changes in face availability across three vertically defined regions of the egocentric video (bottom/middle/top). In early infancy, faces were most common in the middle, with an initial steep decline followed by a modest increase after the first year. In the top region, during the second year, face presence increased and then decreased. The bottom region consistently showed low face presence. These findings suggest that the availability of faces is not only age-dependent but also region-specific, reflecting dynamic reorganisation of everyday visual input. Additionally, face size variability was greater in younger infants, consistent with caregiver-driven interactions. We interpret our findings in the context of emerging motor abilities. By focusing on a specific activity (playtime), this study demonstrates how nuanced patterns can be detected using shorter recordings than in previous studies-enabling scalability and inclusivity of naturalistic research. These results offer new insights into early face availability and demonstrate the value of integrating naturalistic methods with automated analysis to advance developmental theory.