Commits on May 16, 2020

1. Implementations of the custom layers i.e the Correlation layer, Resam… …

   …ple2D layer and the ChannelNorm layer in native Pytorch and C++ to support inference on CPU.
   
   The main bottleneck is the Correlation Layer on which the FlowNetC architecture relies.
   This PR provides 2 implementations of the Correlation layer.
   
   -PyTorch native implementation. This requires no extra setup
   -Optimized C++ implementation for inference on CPU.
   
   Also provided are Pytorch native implementations for Resample2D and Channelnorm.
   Since the Pytorch implementation is quite efficient (compeletely vectorized) with no
   python for loops, C++ implementation is not needed. These layers also work by default
   on the GPU dependening on if the input tensors are on gpu and are slightly slower than
   the provided cuda implementation.
   
   See comments at the top of models.py and networks/FlowNetC.py for more details and
   how to switch to CPU mode.
   
   Backward passes are not yet implemented but will be added in the future.
   
   run_a_pair.py is replaced with a generic script called test.py to simply test functionality.
   run_a_pair.py had hardcoded paths. Also 2 frames from sintel added in test_images dir so that
   functionality and setup can be swiftly checked.
   
   Resolves: NVIDIA#190

   

   zsameem committed May 16, 2020
   Configuration menu

   • View commit details
   • Copy full SHA for 1b038de
   • Browse repository at this point

   Copy the full SHA

   1b038de View commit details

   Browse the repository at this point in the history

2. Revert to use CUDA layers by default

   

   zsameem committed May 16, 2020
   Configuration menu

   • View commit details
   • Copy full SHA for ab10038
   • Browse repository at this point

   Copy the full SHA

   ab10038 View commit details

   Browse the repository at this point in the history