You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
This study aims to show that group equivariant CNNs outperform spatial transformers, on tasks which demand rotation invariance, by providing theoretical background and experimental performance comparison with detailed analysis.
Implementation of STN (Spatial Transformer Network) and ICSTN (Inverse Compositional Spatial Transformer Networks) in Tensorlayer to predict transformation parameters from 2D images.