Learning Question-Guided Video Representation for Multi-Turn Video Question Answering
This work addresses computational inefficiency in video question answering for AI agents interacting with humans, though it is incremental as it builds on existing methods for modality fusion.
The paper tackles the problem of efficiently generating video representations for multi-turn video question answering by proposing a question-guided module that creates token-level video summaries, achieving state-of-the-art performance on the AVSD dataset with improved automatic evaluation metrics.
Understanding and conversing about dynamic scenes is one of the key capabilities of AI agents that navigate the environment and convey useful information to humans. Video question answering is a specific scenario of such AI-human interaction where an agent generates a natural language response to a question regarding the video of a dynamic scene. Incorporating features from multiple modalities, which often provide supplementary information, is one of the challenging aspects of video question answering. Furthermore, a question often concerns only a small segment of the video, hence encoding the entire video sequence using a recurrent neural network is not computationally efficient. Our proposed question-guided video representation module efficiently generates the token-level video summary guided by each word in the question. The learned representations are then fused with the question to generate the answer. Through empirical evaluation on the Audio Visual Scene-aware Dialog (AVSD) dataset, our proposed models in single-turn and multi-turn question answering achieve state-of-the-art performance on several automatic natural language generation evaluation metrics.