Deep visual-semantic alignments for generating image descriptions.
Andrej KarpathyLi Fei-FeiPublished in: CVPR (2015)
Keyphrases
- low level
- high level
- image data
- image representation
- visual appearance
- input image
- image classification
- image content
- single image
- image description
- visual perception
- visual cues
- multiscale
- human visual
- image features
- auto annotation
- low level features
- image retrieval
- image analysis
- segmentation method
- semantic labels
- high level semantics
- human observers
- web images
- image regions
- spatial information
- image segmentation
- spatial relations
- image collections
- edge detection
- test images
- visual information
- visual concepts
- semantically meaningful
- semantic space
- similarity measure
- visually similar
- feature points
- semantic information
- semantic representations
- semantic description
- region of interest
- feature vectors
- image search
- medical images