MaiT: Leverage Attention Masks for More Efficient Image Transformers.
Ling LiAli Shafiee ArdestaniJoseph HassounPublished in: CoRR (2022)
Keyphrases
- input image
- edge detection
- image data
- image content
- image analysis
- multiscale
- image classification
- high resolution
- image retrieval
- image features
- similarity measure
- image pixels
- feature points
- single image
- region of interest
- grey level
- template matching
- test images
- vector field
- image noise
- keypoints
- aerial images
- low level
- image matching
- camera calibration
- hough transform
- natural images
- lighting conditions
- pixel values
- multi view
- bounding box
- image segmentation