Sign in

Improving Maps from CNNs Trained with Sparse, Scribbled Ground Truths Using Fully Connected CRFs.

Luca MaggioloDiego MarcosGabriele MoserDevis Tuia
Published in: IGARSS (2018)
Keyphrases
  • fully connected
  • conditional random fields
  • ground truth
  • scale free
  • high dimensional
  • cellular neural networks
  • image segmentation
  • training set
  • hidden markov models
  • computer vision
  • higher order