Transformers Pay Attention to Convolutions Leveraging Emerging Properties of ViTs by Dual Attention-Image Network.
Yousef YeganehAzade FarshadPeter WeinbergerSeyed-Ahmad AhmadiEhsan AdeliNassir NavabPublished in: ICCV (Workshops) (2023)
Keyphrases
- input image
- image data
- image features
- single image
- image classification
- image analysis
- image content
- high resolution
- image regions
- image collections
- image representation
- image pixels
- image segmentation
- template matching
- test images
- spatial information
- region of interest
- low level
- edge detection
- digital images
- wireless sensor networks
- complex networks
- image retrieval
- image structure
- pixel values
- information redundancy
- auto correlation
- network structure
- vector field
- image matching
- segmentation method
- feature points
- image quality
- medical images
- multiresolution
- feature extraction
- image sequences