FineFormer: Fine-Grained Adaptive Object Transformer for Image Captioning.
Bo WangZhao ZhangJicong FanMingbo ZhaoChoujun ZhanMingliang XuPublished in: ICDM (2022)
Keyphrases
- fine grained
- coarse grained
- input image
- image data
- image regions
- image retrieval
- multiscale
- image features
- keypoints
- lighting conditions
- image content
- spatial relations
- multiple objects
- region of interest
- bounding box
- image segmentation
- image matching
- target object
- tightly coupled
- single image
- feature points
- access control
- spatial relationships
- spatial information
- edge detection
- d objects
- pixel level
- massively parallel
- image sequences
- test images
- distributed systems
- web search
- segmentation method
- object classes
- moving objects
- video sequences
- databases