Object-Centric Unsupervised Image Captioning.
Zihang MengDavid YangXuefei CaoAshish ShahSer-Nam LimPublished in: ECCV (36) (2022)
Keyphrases
- image data
- image features
- image representation
- lighting conditions
- single image
- multiscale
- target object
- high resolution
- image content
- image regions
- normalized correlation
- region of interest
- input image
- low level
- bounding box
- background clutter
- pixel level
- keypoints
- spatial information
- location and orientation
- image analysis
- object features
- complex scenes
- multiple objects
- surface shape
- relative position
- object localization
- unsupervised manner
- image retrieval
- image segmentation
- spatial relationships
- semi supervised
- ground plane
- unsupervised learning
- pixel values
- partial occlusion
- visual appearance
- spatial relations
- image set
- image classification