The faces and hands of caregivers and other social partners offer a rich source of social and causal information that is likely critical for infants' cognitive and linguistic development. Previous work using manual annotation strategies and cross-sectional data has found systematic changes in the proportion of faces and hands in the egocentric perspective of young infants. Here, we validated the use of a modern convolutional neural network (OpenPose) for the detection of faces and hands in naturalistic egocentric videos. We then applied this model to a longitudinal collection of more than 1,700 head-mounted camera videos from three children ages 6 to 32 months. Using these detections, we confirm and extend prior results from cross-sectional studies. First, we found a moderate decrease in the proportion of faces in children's view across age and a higher proportion of hands in view than previously reported. Second, we found variability in the proportion of faces and hands viewed by different children in different locations (e.g., living room vs. kitchen), suggesting that individual activity contexts may shape the social information that infants experience. Third, we found evidence that children may see closer, larger views of people, hands, and faces earlier in development. These longitudinal analyses provide an additional perspective on the changes in the social information in view across the first few years of life and suggest that pose detection models can successfully be applied to naturalistic egocentric video data sets to extract descriptives about infants' changing social environment. (PsycInfo Database Record (c) 2022 APA, all rights reserved).

中文翻译：

---

使用自动检测对婴儿自然视觉体验中的社会信息进行纵向分析。

照顾者和其他社会伙伴的脸和手提供了丰富的社会和因果信息来源，这些信息可能对婴儿的认知和语言发展至关重要。以前使用手动注释策略和横截面数据的工作已经发现，在小婴儿的自我中心视角下，面部和手部的比例发生了系统性变化。在这里，我们验证了使用现代卷积神经网络 (OpenPose) 检测自然主义视频中的面部和手部。然后，我们将该模型应用于纵向收集的 1,700 多个头戴式摄像机视频，这些视频来自三个 6 至 32 个月大的儿童。使用这些检测，我们确认并扩展了横断面研究的先前结果。首先，我们发现儿童的面部比例适度下降 跨年龄段的观点和比以前报道的更高比例的手。其次，我们发现不同儿童在不同地点（例如，客厅与厨房）看到的面孔和手的比例存在差异，这表明个体活动环境可能会影响婴儿体验的社会信息。第三，我们发现有证据表明，儿童在发育早期可能会更近距离、更广阔地观察人、手和脸。这些纵向分析提供了一个关于生命最初几年社会信息变化的额外视角，并表明姿势检测模型可以成功应用于自然主义的以自我为中心的视频数据集，以提取关于婴儿不断变化的社会环境的描述。（PsycInfo 数据库记录 (c) 2022 APA，保留所有权利）。