Hierarchical3D Adapters for Long Video-to-text Summarization
This addresses the problem of generating multi-sentence summaries from long videos for applications like content analysis, but it is incremental as it builds on existing datasets and adapter techniques.
The paper tackles video-to-text summarization for long inputs like hour-long TV shows by extending the SummScreen dataset with full-length videos and using hierarchical adapter modules to incorporate multimodal information efficiently, tuning only 3.8% of parameters and showing superior performance over memory-heavy methods.
In this paper, we focus on video-to-text summarization and investigate how to best utilize multimodal information for summarizing long inputs (e.g., an hour-long TV show) into long outputs (e.g., a multi-sentence summary). We extend SummScreen (Chen et al., 2021), a dialogue summarization dataset consisting of transcripts of TV episodes with reference summaries, and create a multimodal variant by collecting corresponding full-length videos. We incorporate multimodal information into a pre-trained textual summarizer efficiently using adapter modules augmented with a hierarchical structure while tuning only 3.8\% of model parameters. Our experiments demonstrate that multimodal information offers superior performance over more memory-heavy and fully fine-tuned textual summarization methods.