A Video-grounded Dialogue Dataset and Metric for Event-driven Activities
This addresses the need for better datasets and metrics in video-grounded dialogue for event-driven activities, though it is incremental as it builds on existing work.
The authors introduced VDAct, a dataset of 3,000 dialogues with over 30,000 question-answer pairs from 1,000 videos for video-grounded dialogue on event-driven activities, and VDEval, a session-based evaluation metric that shows higher correlation with human assessments than existing metrics.
This paper presents VDAct, a dataset for a Video-grounded Dialogue on Event-driven Activities, alongside VDEval, a session-based context evaluation metric specially designed for the task. Unlike existing datasets, VDAct includes longer and more complex video sequences that depict a variety of event-driven activities that require advanced contextual understanding for accurate response generation. The dataset comprises 3,000 dialogues with over 30,000 question-and-answer pairs, derived from 1,000 videos with diverse activity scenarios. VDAct displays a notably challenging characteristic due to its broad spectrum of activity scenarios and wide range of question types. Empirical studies on state-of-the-art vision foundation models highlight their limitations in addressing certain question types on our dataset. Furthermore, VDEval, which integrates dialogue session history and video content summaries extracted from our supplementary Knowledge Graphs to evaluate individual responses, demonstrates a significantly higher correlation with human assessments on the VDAct dataset than existing evaluation metrics that rely solely on the context of single dialogue turns.