Video Scene Graph Generation (VidSGG) is a research field focused on converting visual input from video streams into structured knowledge representations.
These scene graphs contain nodes (objects) and edges (relationships) that help machines understand context and interactions within each frame over time.

Recent Publications
- HIG: Hierarchical Interlacement Graph Approach to Scene Graph Generation in Video Understanding (CVPR 2024) [PDF] Demo
- CYCLO: Cyclic Graph Transformer Approach to Multi-Object Relationship Modeling in Aerial Videos (NeurIPS, 2024) [PDF] Demo
- HyperGLM: HyperGraph for Video Scene Graph Generation and Anticipation (arXiv, 2025) [PDF] Demo