Self-Annotated Training for Controllable Image Captioning.
Zhangzi ZhuTianlei WangHong QuPublished in: CoRR (2021)
Keyphrases
- input image
- image features
- image content
- image data
- multiscale
- image collections
- image pixels
- template matching
- image segmentation
- similarity measure
- image analysis
- edge detection
- image classification
- test images
- image regions
- single image
- feature points
- image structure
- region of interest
- keypoints
- image noise
- object detectors
- grey level
- training samples
- vector field
- energy function
- segmentation algorithm
- neural network
- image representation
- low level
- high resolution
- image retrieval
- computer vision