A weight-learning network is trained end-to-end with a base network to dynamically modulate its weights from the parameters of classical image operators, with an efficient single-layer variant that shares most computation.
Fast guided filter,
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2019 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
A General Decoupled Learning Framework for Parameterized Image Operators
A weight-learning network is trained end-to-end with a base network to dynamically modulate its weights from the parameters of classical image operators, with an efficient single-layer variant that shares most computation.