Diverse Image Captioning with Context-Object Split Latent Spaces.
Shweta MahajanStefan RothPublished in: NeurIPS (2020)
Keyphrases
- input image
- image features
- multiscale
- image regions
- keypoints
- image data
- image analysis
- image retrieval
- image content
- spatial relationships
- image representation
- visual context
- image segmentation
- lighting conditions
- spatial context
- similar objects
- high resolution
- multiple objects
- complex scenes
- region of interest
- single image
- contextual information
- spatial relations
- image classification
- bounding box
- d objects
- target object
- object localization
- normalized correlation
- pixel level
- partial occlusion
- image set
- latent variables
- location and orientation
- surface shape
- spatial information
- object shape
- test images
- background clutter
- three dimensional objects
- visual appearance
- image matching
- foreground and background
- image segments
- object models
- segmentation method
- boundary contour
- multi view