Generative Compositional Augmentations for Scene Graph Prediction.
Boris KnyazevHarm de VriesCatalina CangeaGraham W. TaylorAaron C. CourvilleEugene BelilovskyPublished in: ICCV (2021)
Keyphrases
- graph theory
- prediction accuracy
- d scene
- weighted graph
- generative model
- video sequences
- graph representation
- graph structure
- prediction algorithm
- random walk
- protein function prediction
- complex scenes
- scene analysis
- prediction model
- directed graph
- connected components
- input image
- moving objects
- three dimensional
- graph theoretic
- image sequences
- multiple images
- prediction error
- real scenes
- graph mining
- scene classification
- observed scene
- bipartite graph
- scene understanding
- spanning tree
- geometric constraints
- computer vision
- dynamic scenes
- image classification
- high resolution
- optical flow
- object recognition
- image segmentation