Video Dataset for Recognition of Family Dining Behaviors and Social Interaction Patterns

With the prevalence of fast-paced lifestyles in society, people often overlook the importance of family dining time, which not only affects emotional communication among family members but also has a negative impact on the development of children's social skills. Current market solutions are mostly limited to offline observation, which is costly and has limited samples, thus failing to provide sufficient empirical support for research. This dataset aims to address the issue of insufficient emotional communication within families by recognizing dining behaviors and social interactions in videos, and provides data support for social behavior modeling. Data collection is conducted in real family environments, using high-definition cameras and 360-degree panoramic shooting to ensure detailed capture. The data is checked for consistency through multiple rounds of manual proofreading and is reviewed and annotated by experts with backgrounds in psychology and sociology to ensure high quality. The annotation team consists of 10 professionals from related fields, and during the data preprocessing stage, videos undergo complex processes such as denoising, segmentation, and behavior annotation, and are categorically stored in MP4 format for easy retrieval and reuse.