Boosting global scene classification accuracy by discriminative region localization.
Thanh Duc NgoDuy-Dinh LeShin'ichi SatohPublished in: ICIP (2011)
Keyphrases
- classification accuracy
- feature selection
- precise localization
- input image
- improving classification accuracy
- d scene
- image regions
- scene representation
- global image features
- video sequences
- bayesian network classifiers
- image sequences
- feature space
- feature extraction
- machine learning
- multiple images
- scene classification
- semi supervised
- feature set
- single image
- observed scene
- dynamic scenes
- discriminative features
- region of interest
- conditional likelihood
- outdoor scenes
- scene understanding
- ensemble learning
- ensemble methods
- feature subset
- optical flow
- support vector
- computer vision
- learning algorithm