Streaming Video Model
This work addresses the need for a unified architecture in video understanding, potentially benefiting researchers and practitioners in computer vision, though it appears incremental as it builds on existing transformer and video modeling concepts.
The paper tackles the problem of unifying video understanding tasks, which traditionally use separate architectures for sequence-based and frame-based tasks, by proposing a streaming video architecture called Streaming Vision Transformer (S-ViT), achieving state-of-the-art accuracy in action recognition and competitive advantage in multiple object tracking.
Video understanding tasks have traditionally been modeled by two separate architectures, specially tailored for two distinct tasks. Sequence-based video tasks, such as action recognition, use a video backbone to directly extract spatiotemporal features, while frame-based video tasks, such as multiple object tracking (MOT), rely on single fixed-image backbone to extract spatial features. In contrast, we propose to unify video understanding tasks into one novel streaming video architecture, referred to as Streaming Vision Transformer (S-ViT). S-ViT first produces frame-level features with a memory-enabled temporally-aware spatial encoder to serve the frame-based video tasks. Then the frame features are input into a task-related temporal decoder to obtain spatiotemporal features for sequence-based tasks. The efficiency and efficacy of S-ViT is demonstrated by the state-of-the-art accuracy in the sequence-based action recognition task and the competitive advantage over conventional architecture in the frame-based MOT task. We believe that the concept of streaming video model and the implementation of S-ViT are solid steps towards a unified deep learning architecture for video understanding. Code will be available at https://github.com/yuzhms/Streaming-Video-Model.