Object Counts! Bringing Explicit Detections Back into Image Captioning.
Josiah WangPranava Swaroop MadhyasthaLucia SpeciaPublished in: NAACL-HLT (2018)
Keyphrases
- image data
- single image
- image classification
- image features
- image regions
- lighting conditions
- input image
- visual appearance
- bounding box
- target object
- region of interest
- image content
- multiscale
- complex scenes
- image analysis
- normalized correlation
- object localization
- pixel level
- keypoints
- image representation
- image segmentation
- multiple objects
- spatial relations
- image set
- test images
- similar objects
- pixel values
- spatial information
- edge detection
- image retrieval
- feature points
- d objects
- location and orientation
- object features
- object representation
- foreground and background
- background clutter
- object models
- moving objects
- spatial relationships
- high resolution
- low level
- relative position
- segmentation method
- image quality
- segmentation algorithm
- object shape
- light source
- image matching