Look Twice as Much as You Say: Scene Graph Contrastive Learning for Self-Supervised Image Caption Generation.
Chunhui ZhangChao HuangYouhuan LiXiangliang ZhangYanfang YeChuxu ZhangPublished in: CIKM (2022)
Keyphrases
- input image
- single image
- multiscale
- image data
- learning algorithm
- image regions
- image representation
- image set
- image content
- image classification
- low level
- image sequences
- geometric information
- real world scenes
- image retrieval
- ground plane
- complex scenes
- geometric constraints
- outdoor scenes
- scene classification
- piecewise planar
- lighting conditions
- test images
- d scene
- spatial information
- segmentation algorithm
- feature points
- image features
- three dimensional