A Two Layer Pairwise Framework to Approximate Superpixel-Based Higher-Order Conditional Random Field for Semantic Segmentation.
Li SulimowiczIshfaq AhmadAlexander J. AvedPublished in: ICME Workshops (2018)
Keyphrases
- conditional random fields
- semantic segmentation
- pairwise
- higher order
- superpixels
- markov random field
- probabilistic model
- crf model
- long range
- natural images
- multi class
- information extraction
- graphical models
- hidden markov models
- scene classification
- generative model
- weakly supervised
- machine learning
- object class
- bayesian networks
- segmentation method
- semi supervised
- natural scenes
- object recognition
- learning algorithm