IntenseWebs/openvino
3
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

image-fix (#13867)

Browse Source
This commit is contained in:
Maciej Smyk 2022-12-12 13:04:57 +01:00 committed by GitHub
parent ce5c0ff1dc
commit 59ea1c43c4
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
9 changed files with 7129 additions and 16 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

3
docs/MO_DG/img/optimizations/groups.png
View File

@ -1,3 +0,0 @@
version https://git-lfs.github.com/spec/v1
oid sha256:3812efef32bd7f1bf40b130d5d522bc3df6aebd406bd1186699d214bca856722
size 43721

1117
docs/MO_DG/img/optimizations/groups.svg Normal file
View File

File diff suppressed because it is too large Load Diff
Side by Side

After

Width:  |  Height:  |  Size: 139 KiB

3
docs/MO_DG/img/optimizations/inception_v4.png
View File

@ -1,3 +0,0 @@
version https://git-lfs.github.com/spec/v1
oid sha256:0e232c47e8500f42bd0e1f2b93f94f58e2d59caee149c687be3cdc3e8a5be59a
size 18417

1774
docs/MO_DG/img/optimizations/inception_v4.svg Normal file
View File

File diff suppressed because it is too large Load Diff
Side by Side

After

Width:  |  Height:  |  Size: 170 KiB

3
docs/MO_DG/img/optimizations/resnet_269.png
View File

@ -1,3 +0,0 @@
version https://git-lfs.github.com/spec/v1
oid sha256:92d36b9527a3e316cd9eb2b6f5054c312466df004e4aa9c3458e165330bc6561
size 24157

1696
docs/MO_DG/img/optimizations/resnet_269.svg Normal file
View File

File diff suppressed because it is too large Load Diff
Side by Side

After

Width:  |  Height:  |  Size: 344 KiB

3
docs/MO_DG/img/optimizations/resnet_optimization.png
View File

@ -1,3 +0,0 @@
version https://git-lfs.github.com/spec/v1
oid sha256:2adeca1e3512b9fe7b088a5412ce21592977a1f352a013735537ec92e895dc94
size 15653

2538
docs/MO_DG/img/optimizations/resnet_optimization.svg Normal file
View File

File diff suppressed because it is too large Load Diff
Side by Side

After

Width:  |  Height:  |  Size: 486 KiB

8
docs/MO_DG/prepare_model/Model_Optimization_Techniques.md
View File

@ -26,7 +26,7 @@ This optimization method consists of three stages:
The picture below shows the depicted part of Caffe Resnet269 topology where `BatchNorm` and `ScaleShift` layers will be fused to `Convolution` layers. The picture below shows the depicted part of Caffe Resnet269 topology where `BatchNorm` and `ScaleShift` layers will be fused to `Convolution` layers.
![Caffe ResNet269 block before and after optimization generated with Netscope*](../img/optimizations/resnet_269.png) ![Caffe ResNet269 block before and after optimization generated with Netscope*](../img/optimizations/resnet_269.svg)
* * * * * *
@ -38,7 +38,7 @@ ResNet optimization is a specific optimization that applies to Caffe ResNet topo
In the picture below, you can see the original and optimized parts of a Caffe ResNet50 model. The main idea of this optimization is to move the stride that is greater than 1 from Convolution layers with the kernel size = 1 to upper Convolution layers. In addition, the Model Optimizer adds a Pooling layer to align the input shape for a Eltwise layer, if it was changed during the optimization. In the picture below, you can see the original and optimized parts of a Caffe ResNet50 model. The main idea of this optimization is to move the stride that is greater than 1 from Convolution layers with the kernel size = 1 to upper Convolution layers. In addition, the Model Optimizer adds a Pooling layer to align the input shape for a Eltwise layer, if it was changed during the optimization.
![ResNet50 blocks (original and optimized) from Netscope](../img/optimizations/resnet_optimization.png) ![ResNet50 blocks (original and optimized) from Netscope](../img/optimizations/resnet_optimization.svg)
In this example, the stride from the `res3a_branch1` and `res3a_branch2a` Convolution layers moves to the `res2c_branch2b` Convolution layer. In addition, to align the input shape for `res2c` Eltwise, the optimization inserts the Pooling layer with kernel size = 1 and stride = 2. In this example, the stride from the `res3a_branch1` and `res3a_branch2a` Convolution layers moves to the `res2c_branch2b` Convolution layer. In addition, to align the input shape for `res2c` Eltwise, the optimization inserts the Pooling layer with kernel size = 1 and stride = 2.
@ -48,7 +48,7 @@ In this example, the stride from the `res3a_branch1` and `res3a_branch2a` Convol
Grouped convolution fusing is a specific optimization that applies for TensorFlow topologies. The main idea of this optimization is to combine convolutions results for the `Split` outputs and then recombine them using `Concat` operation in the same order as they were out from `Split`. Grouped convolution fusing is a specific optimization that applies for TensorFlow topologies. The main idea of this optimization is to combine convolutions results for the `Split` outputs and then recombine them using `Concat` operation in the same order as they were out from `Split`.
![Split→Convolutions→Concat block from TensorBoard*](../img/optimizations/groups.png) ![Split→Convolutions→Concat block from TensorBoard*](../img/optimizations/groups.svg)
* * * * * *
@ -62,4 +62,4 @@ On the picture below you can see two visualized Intermediate Representations (IR
The first one is original IR that will be produced by the Model Optimizer. The first one is original IR that will be produced by the Model Optimizer.
The second one will be produced by the Model Optimizer with key `--finegrain_fusing InceptionV4/InceptionV4/Conv2d_1a_3x3/Conv2D`, where you can see that `Convolution` was not fused with `Mul1_3752` and `Mul1_4061/Fused_Mul_5096/FusedScaleShift_5987` operations. The second one will be produced by the Model Optimizer with key `--finegrain_fusing InceptionV4/InceptionV4/Conv2d_1a_3x3/Conv2D`, where you can see that `Convolution` was not fused with `Mul1_3752` and `Mul1_4061/Fused_Mul_5096/FusedScaleShift_5987` operations.
![TF InceptionV4 block without/with key --finegrain_fusing (from IR visualizer)](../img/optimizations/inception_v4.png) ![TF InceptionV4 block without/with key --finegrain_fusing (from IR visualizer)](../img/optimizations/inception_v4.svg)