Scene Retrieval from Multiple Resolution Generated Images Based on Text-to-Image GAN.
Rintaro YanagiRen TogoTakahiro OgawaMiki HaseyamaPublished in: ISCAS (2019)
Keyphrases
- input image
- image retrieval
- image regions
- scene images
- image data
- image collections
- multiple objects
- image set
- text queries
- visual appearance
- image features
- reference images
- image description
- test images
- imaging process
- web images
- text extraction
- single image
- multiple images
- image segments
- synthesized images
- trademark images
- dense depth map
- high resolution
- image resolution
- complex scenes
- visual and textual features
- higher resolution
- image classification
- image analysis
- image content
- acquired images
- image database
- high spatial resolution
- vanishing points
- edge detection
- image search
- textual descriptions
- relevant images
- calibrated cameras
- visual features
- higher spatial resolution
- textual query
- image matching
- intensity images
- piecewise planar
- visual similarity
- low resolution
- geometric constraints
- image pyramid
- segmentation method
- illumination conditions
- combining information from multiple
- fundamental problem in computer vision
- relevance feedback
- depth map
- scene geometry
- d scene
- lighting conditions
- pixel values
- image representation
- handwritten documents
- defocused images
- feature points
- textual and visual information
- complex background
- historical manuscripts
- image processing
- image sequences
- image quality
- super resolution
- colour constancy
- gray level
- text regions