Compositionality, MDL Priors, and Object Recognition.
Elie BienenstockStuart GemanDaniel PotterPublished in: NIPS (1996)
Keyphrases
- object recognition
- model selection
- minimum description length
- natural images
- object detection
- image features
- computer vision
- scene classification
- object class
- d objects
- visual recognition
- image understanding
- bayesian framework
- mdl principle
- object categorization
- recognition scheme
- object categories
- visual object recognition
- constrained search
- penalized likelihood
- feature detection
- keypoints
- object classification
- maximum a posteriori
- image matching
- biologically motivated
- image representation
- model based object recognition
- graph matching
- template matching
- prior probabilities
- information criterion
- maximum likelihood
- image sequences