commit fe95d7203ba7a6ac31e0656190ba56de6fcf4735
author Diego Lopez Recas <Diego.LopezRecas@arm.com> Fri Mar 19 12:40:16 2021 +0000
committer Colm Donelan <colm.donelan@arm.com> Wed Apr 21 14:51:17 2021 +0000
tree 4b127a99c97364d6b9062d93ece26ccafa60579d
parent 7612bd6cc385dfbf54f831a6349f3a9363c6d0a2

Fold PAD into Pooling2d if possible

Some models would add a PAD layer before a pooling when they can't
express their padding configuration as SAME or VALID.

ArmNN can potentially merge the two merge the two because pooling
layers are described with explicit padding.

The merge is possible if the extra padding is neutral in the combined
pooling operation. A merged operation can only fuse paddings in the
dimensions that accept explicit padding in a pooling operation, i.e. the
spatial dimensions.

Signed-off-by: Diego Lopez Recas <diego.lopez.recas@gmail.com>
Signed-off-by: Colm Donelan <Colm.Donelan@arm.com>

Change-Id: Icd54718dcd9e797c923456b7fa6e0213e288e668

src/armnn/Network.cpp

diff --git a/src/armnn/Network.cpp b/src/armnn/Network.cpp
index 860048f..c99690e 100644
--- a/src/armnn/Network.cpp
+++ b/src/armnn/Network.cpp
@@ -1563,6 +1563,7 @@
                                                 TransposeAsReshape(),
                                                 OptimizeConsecutiveReshapes(),
                                                 FoldPadIntoConvolution2d(),
+                                                FoldPadIntoPooling2d(),
                                                 PermuteAndBatchToSpaceAsDepthToSpace(),
                                                 TransposeAndBatchToSpaceAsDepthToSpace(),
                                                 FuseBatchNormIntoConvolution2DFloat32(),