ConIS: controllable text-driven image stylization with semantic intensity.
Gaoming YangChanggeng LiJi ZhangPublished in: Multim. Syst. (2024)
Keyphrases
- single image
- input image
- intensity values
- image features
- image data
- image analysis
- multiscale
- image classification
- intensity variations
- image content
- semantic information
- feature points
- image representation
- test images
- low level
- image retrieval
- segmentation method
- intensity images
- image synthesis
- intensity levels
- web images
- scanned documents
- high level
- semantic network
- image structure
- pixel values
- image pixels
- segmentation algorithm
- semantic content
- image quality
- semantic representation
- pixel intensities
- edge detection
- high resolution
- keywords
- illumination change
- image segmentation
- arabic text