Query-based Image Captioning from Multi-context 360cdegree Images.
Koki MaedaShuhei KuritaTaiki MiyanishiNaoaki OkazakiPublished in: EMNLP (Findings) (2023)
Keyphrases
- input image
- image data
- image collections
- test images
- image retrieval
- target images
- relevant images
- image features
- visually similar
- original images
- retrieved images
- image content
- image similarity
- web images
- image classification
- street view
- image pixels
- image regions
- pixel values
- object retrieval
- million images
- high contrast
- image search
- image analysis
- web image search
- retrieving images
- region of interest
- image database
- intensity values
- image set
- segmentation method
- image matching
- image dataset
- image processing algorithms
- image noise
- image structure
- reference images
- edge detection
- lighting conditions
- semantic gap
- normalized correlation
- sample images
- imaging process
- relevance feedback
- synthesized images
- segmented images
- text queries
- grey level
- digital imaging
- labeled images
- image annotation
- visual appearance
- keypoints
- semantic meaning
- segmentation algorithm
- feature points
- false matches
- illumination conditions
- textual descriptions
- contrast enhancement
- image segmentation
- gray value
- visual features
- ccd camera
- visual concepts
- cbir systems
- aerial images
- image registration
- keywords
- image processing
- image restoration
- image representation
- image quality
- object recognition
- multiscale
- face recognition