A Local-to-Global Approach to Multi-modal Movie Scene Segmentation
This addresses movie understanding for video analysis applications, though it appears incremental as it builds on existing scene segmentation work with a new dataset and hierarchical framework.
The paper tackles movie scene segmentation by building a large-scale dataset (MovieScenes with 21K annotated segments from 150 movies) and proposing a local-to-global multi-modal framework that integrates information across clip, segment, and movie levels, consistently outperforming previous methods and showing that pretraining on MovieScenes improves existing approaches.
Scene, as the crucial unit of storytelling in movies, contains complex activities of actors and their interactions in a physical environment. Identifying the composition of scenes serves as a critical step towards semantic understanding of movies. This is very challenging -- compared to the videos studied in conventional vision problems, e.g. action recognition, as scenes in movies usually contain much richer temporal structures and more complex semantic information. Towards this goal, we scale up the scene segmentation task by building a large-scale video dataset MovieScenes, which contains 21K annotated scene segments from 150 movies. We further propose a local-to-global scene segmentation framework, which integrates multi-modal information across three levels, i.e. clip, segment, and movie. This framework is able to distill complex semantics from hierarchical temporal structures over a long movie, providing top-down guidance for scene segmentation. Our experiments show that the proposed network is able to segment a movie into scenes with high accuracy, consistently outperforming previous methods. We also found that pretraining on our MovieScenes can bring significant improvements to the existing approaches.