Feature Deformation Meta-Networks in Image Captioning of Novel Objects.
Tingjia CaoKe HanXiaomei WangLin MaYanwei FuYu-Gang JiangXiangyang XuePublished in: AAAI (2020)
Keyphrases
- image features
- feature values
- multiple objects
- multiscale
- complex scenes
- image pixels
- image regions
- image representation
- image data
- input image
- object features
- keypoints
- image retrieval
- geometrical features
- single image
- spatial relations
- image analysis
- image classification
- image content
- feature points
- object shape
- image processing
- image segments
- deformable objects
- object shapes
- segmented regions
- scale invariant features
- image segmentation
- target object
- spatial relationships
- geometric constraints
- spatial information
- edge detection
- vector field
- high resolution
- image deformations
- image registration
- linear features
- spatial configuration
- rigid transformation
- registration process
- cluttered scenes
- similar objects
- bounding box
- pixel level