A framework for continuous fingerspelling spotting for H.264/AVC compressed videos using spatio-temporal Markov random field.
Anjan Kumar TalukdarManas Kamal BhuyanPublished in: Multim. Tools Appl. (2021)
Keyphrases
- markov random field
- spatio temporal
- higher order
- image segmentation
- mrf model
- random fields
- parameter estimation
- maximum a posteriori
- graph cuts
- belief propagation
- discriminative random fields
- image restoration
- energy function
- energy minimization
- message passing
- sign language
- potential functions
- markov random field model
- pairwise
- maximum likelihood
- moving objects
- loopy belief propagation