FactorNet: Holistic Actor, Object, and Scene Factorization for Action Recognition in Videos.
Nitika NigamTanima DuttaHari Prabhat GuptaPublished in: IEEE Trans. Circuits Syst. Video Technol. (2022)
Keyphrases
- multiple objects
- moving objects
- complex scenes
- action recognition in videos
- real world scenes
- object models
- action recognition
- target object
- spatial relations
- uncalibrated images
- d objects
- d scene
- object motion
- viewing position
- multiple images
- object tracking
- three dimensional
- affine camera
- video sequences
- real world objects
- video scene
- real objects
- geometric information
- input image
- viewpoint
- ground plane
- dynamic scenes
- single image
- object model
- rigid objects
- image sequences
- background pixels
- machine learning
- uncalibrated cameras
- viewing direction
- camera positions
- relative position
- image regions
- camera motion
- object segmentation