Fully Convolutional Video Captioning with Coarse-to-Fine and Inherited Attention.
Kuncheng FangLian ZhouCheng JinYuejie ZhangKangnian WengTao ZhangWeiguo FanPublished in: AAAI (2019)
Keyphrases
- coarse to fine
- successive frames
- multiscale
- multiresolution
- image registration
- hierarchical segmentation
- object detection
- video content
- hierarchical representation
- convolutional network
- dynamic programming
- optical flow estimation
- video data
- active shape model
- video frames
- video sequences
- matching scheme
- natural images
- feature correspondences
- training data
- visual attention
- deformable surface model
- facial landmarks
- computer vision
- image processing