Coarse-to-Fine Dual Encoders are Better Frame Identification Learners.
Kaikai AnCe ZhengBofei GaoHaozhe ZhaoBaobao ChangPublished in: EMNLP (Findings) (2023)
Keyphrases
- coarse to fine
- successive frames
- multiscale
- multiresolution
- feature correspondences
- hierarchical segmentation
- image registration
- object detection
- hierarchical representation
- e learning
- learning environment
- active shape model
- dynamic programming
- matching scheme
- video frames
- image pyramid
- video compression
- reinforcement learning
- machine learning