Bottom-Up and Top-Down Attention for Image Captioning and VQA.
Peter AndersonXiaodong HeChris BuehlerDamien TeneyMark JohnsonStephen GouldLei ZhangPublished in: CoRR (2017)
Keyphrases
- image features
- image content
- single image
- input image
- image data
- image segmentation
- image database
- image retrieval
- image representation
- low level
- multiscale
- high resolution
- region of interest
- image classification
- segmentation method
- image analysis
- image regions
- spatial information
- pixel level
- segmentation algorithm
- edge detection
- multiresolution
- test images
- high quality
- image pixels
- computer vision