ER-SAN: Enhanced-Adaptive Relation Self-Attention Network for Image Captioning.
Jingyu LiZhendong MaoShancheng FangHao LiPublished in: IJCAI (2022)
Keyphrases
- multiscale
- input image
- image classification
- image features
- template matching
- image data
- single image
- image content
- image representation
- wireless sensor networks
- image analysis
- image retrieval
- edge detection
- image regions
- object oriented
- image collections
- image processing
- pixel values
- segmentation algorithm
- network structure
- test images
- network traffic
- image pixels
- complex networks
- natural images
- image matching
- keypoints
- segmentation method
- feature points
- moving objects
- video sequences
- similarity measure
- image segmentation