A teacher-loss confidence weighting for knowledge distillation lets a 93-percent-smaller pose-regression network nearly match the teacher's visual-odometry accuracy.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2019 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Distilling Knowledge From a Deep Pose Regressor Network
A teacher-loss confidence weighting for knowledge distillation lets a 93-percent-smaller pose-regression network nearly match the teacher's visual-odometry accuracy.