diff options
| author | Diego Lopez Recas <Diego.LopezRecas@arm.com> | 2021-03-19 12:40:16 +0000 |
|---|---|---|
| committer | Diego Lopez Recas <diego.lopez.recas@gmail.com> | 2021-03-22 10:35:04 +0000 |
| commit | 51ce7d487c761358de105f82ff90553570aedac0 (patch) | |
| tree | 763e1380249af2dee57e98b022b915c9cae8a1ee /src/armnn/Network.cpp | |
| parent | 8b35d0922e040ecf4d6b2f5012af356c383f5561 (diff) | |
| download | armnn-51ce7d487c761358de105f82ff90553570aedac0.tar.gz | |
Fold PAD into Pooling2d
Some models would add a PAD layer before a pooling when they can't
express their padding configuration as SAME or VALID.
Arm NN can merge the two because pooling layers are described with
explicit padding.
Signed-off-by: Diego Lopez Recas <diego.lopez.recas@gmail.com>
Change-Id: Id048186db6a005e0257bfbc1406c3b0dab2cdd58
Diffstat (limited to 'src/armnn/Network.cpp')
| -rw-r--r-- | src/armnn/Network.cpp | 1 |
1 files changed, 1 insertions, 0 deletions
diff --git a/src/armnn/Network.cpp b/src/armnn/Network.cpp index 9373a6ac15..8dad3bba56 100644 --- a/src/armnn/Network.cpp +++ b/src/armnn/Network.cpp @@ -1544,6 +1544,7 @@ IOptimizedNetworkPtr Optimize(const INetwork& inNetwork, TransposeAsReshape(), OptimizeConsecutiveReshapes(), FoldPadIntoConvolution2d(), + FoldPadIntoPooling2d(), PermuteAndBatchToSpaceAsDepthToSpace(), TransposeAndBatchToSpaceAsDepthToSpace(), FuseBatchNormIntoConvolution2DFloat32(), |