Generating Descriptions for Sequential Images with Local-Object Attention and Global Semantic Context Modelling.
Jing SuChenghua LinMian ZhouQingyun DaiHaoyu LvPublished in: CoRR (2020)
Keyphrases
- image database
- target object
- image data
- lighting conditions
- input image
- image retrieval
- test images
- image regions
- multiple objects
- semantic context
- image set
- image matching
- region of interest
- image features
- edge detection
- d objects
- object recognition
- image collections
- spatial relationships
- bounding box
- complex scenes
- partial occlusion
- high level
- computer vision
- multiscale
- high level semantics