Effects of variability in synthetic training data on convolutional neural networks for 3D head reconstruction.
Jan Philip GöpfertChristina GöpfertMario BotschBarbara HammerPublished in: SSCI (2017)
Keyphrases
- convolutional neural networks
- training data
- convolutional network
- data sets
- training set
- decision trees
- three dimensional
- domain knowledge
- image reconstruction
- real world
- test data
- learning algorithm
- classification accuracy
- prior knowledge
- software product line
- head pose estimation
- reconstruction process
- classification models
- machine learning
- discrete tomography
- training instances
- training dataset
- training process
- support vector machine
- noisy data
- training examples
- training samples
- supervised learning
- image registration
- semi supervised