A Semisupervised CRF Model for CNN-Based Semantic Segmentation With Sparse Ground Truth.
Luca MaggioloDiego MarcosGabriele MoserSebastiano B. SerpicoDevis TuiaPublished in: IEEE Trans. Geosci. Remote. Sens. (2022)
Keyphrases
- semantic segmentation
- conditional random fields
- crf model
- ground truth
- semi supervised
- superpixels
- generative model
- information extraction
- graphical models
- higher order
- probabilistic model
- hidden markov models
- long range
- pairwise
- markov random field
- maximum entropy
- test images
- sparse representation
- natural language
- multiscale
- image segmentation
- natural images
- weakly supervised