Rethinking Image-to-Video Adaptation: An Object-centric Perspective.
Rui QianShuangrui DingDahua LinPublished in: CoRR (2024)
Keyphrases
- image frames
- image features
- image regions
- single image
- image data
- image classification
- weakly labeled
- input image
- target object
- object motion
- region of interest
- pixel level
- complex scenes
- multiple objects
- static images
- lighting conditions
- temporal continuity
- multiscale
- image representation
- spatial relations
- video sequences
- low level
- normalized correlation
- spatial relationships
- keypoints
- image content
- edge detection
- bounding box
- video data
- successive frames
- combining information from multiple
- object shapes
- raw image
- camera movement
- video content
- key frames
- test images
- d objects
- image retrieval
- object models
- image set
- pre trained
- object tracking
- feature points
- viewpoint