Style-transfer GANs for bridging the domain gap in synthetic pose estimator training.
Pavel RojtbergThomas PöllabauerArjan KuijperPublished in: CoRR (2020)
Keyphrases
- cross domain
- transfer learning
- d objects
- real world
- supervised learning
- domain specific
- least squares
- training phase
- pose estimation
- online learning
- training samples
- real images are presented
- neural network
- real scenes
- training algorithm
- training process
- partial occlusion
- domain independent
- training examples
- feature points
- multi view