Towards Scalable and Ecological Methods for Early Social Development Research: Validating Automated Facial Expression Estimation in Egocentric Video

Wearable head-mounted cameras offer a window into the infant’s social world, yet manual coding remains a computational bottleneck. To address this limitation, we validate a scalable, open-source pipeline for automated facial expression estimation using the EfficientNet-B0 architecture on naturalistic, egocentric videos. Despite non-canonical viewpoints, poor lighting and motion artifacts, the model achieves performance (F1 = 0.543) comparable to state-of-the-art benchmarks. Crucially, we demonstrate high representational alignment (r = 0.76) between model confusion patterns and human perception, suggesting that the architecture serves as a faithful proxy for the adult social-perceptual system. While categorical accuracy varies, the model’s distributional probabilities capture the graded nature of affect. We also identify a critical divergence in neutral expression processing, where model and human ambiguity dissociate (r = -0.26). These findings establish a robust, high-throughput instrument for quantifying the social input infants receive, enabling a large-scale, "big data" science of early social development.