Remote Meeting Attendee Attention State Information Extraction Video Dataset

With the increasing prevalence of remote work, ensuring the effectiveness of meetings has become a significant challenge. Traditional methods of evaluating meeting effectiveness rely on subjective feedback, making it difficult to conduct precise analysis through data-driven approaches. This poses a dilemma for meeting organizers. The construction of this dataset aims to provide a quantifiable and more objective evaluation method to help identify the attention state of attendees during meetings. Data collection employs high-definition camera equipment set in various office environments to simulate real remote meeting scenarios. Quality control is ensured through multiple rounds of annotation and consistency checks, carried out by a team of 10 individuals with professional backgrounds in behavior analysis. Data preprocessing includes keyframe extraction, feature point marking, and time series slicing, finally stored and organized into subsets of different meeting scenarios in MP4 format.