Social Group Activity Recognition

Social group activity recognition aims to automatically identify the actions of groups of people and their individual members within video data. Current research heavily utilizes transformer-based architectures, often employing self-supervised learning techniques to leverage unlabeled video data and effectively model long-range spatiotemporal dependencies within group interactions. These advancements focus on improving feature extraction by incorporating attention mechanisms to better capture the complex relationships between individuals and their collective actions. This field holds significant potential for applications in areas such as video surveillance, human-computer interaction, and sports analytics.