Normalized and Geometry-Aware Self-Attention Network for Image Captioning.
Longteng GuoJing LiuXinxin ZhuPeng YaoShichen LuHanqing LuPublished in: CoRR (2020)
Keyphrases
- input image
- image analysis
- image features
- single image
- image data
- multiscale
- template matching
- image content
- three dimensional
- image classification
- image pixels
- image retrieval
- similarity measure
- low level
- image collections
- geometric features
- normalized correlation
- pixel values
- image set
- lighting conditions
- energy function
- region of interest
- network structure
- image representation
- wireless sensor networks
- high resolution
- image regions
- edge detection
- geometric constraints
- geometric structure
- relative position
- optical flow
- geometric information
- image sequences