diff options
| author | Finn Williams <Finn.Williams@arm.com> | 2020-07-23 12:55:12 +0100 |
|---|---|---|
| committer | Teresa Charlin <teresa.charlinreyes@arm.com> | 2020-08-12 17:06:37 +0100 |
| commit | a4983cec09a3e24bf4e99abd31aa11842e8b365f (patch) | |
| tree | 0a09be7cdad7ad5c15bb24c67abf0f4cd4d6e47e /ModelToINetworkConverter.cpp | |
| parent | bf866e2dc7bd5936788fe213b5c0f74483ec1532 (diff) | |
| download | android-nn-driver-a4983cec09a3e24bf4e99abd31aa11842e8b365f.tar.gz | |
IVGCVSW-4931 Update NN Driver to support dynamic tensors
* Change NN Driver m_Network to now have ShapeInferenceMethod::InferAndValidate
* Implement dynamic tensor support for:
- ArgMinMax layer
- Pooling2d layer
- Activation layer
* Skip dynamic tensor tests for any HAL other than 1.3
Change-Id: Icf66c968e49cdd4822b8c79c5f18b3f9e97dc53f
Signed-off-by: Finn Williams <Finn.Williams@Arm.com>
Signed-off-by: Teresa Charlin <teresa.charlinreyes@arm.com>
Diffstat (limited to 'ModelToINetworkConverter.cpp')
| -rw-r--r-- | ModelToINetworkConverter.cpp | 11 |
1 files changed, 10 insertions, 1 deletions
diff --git a/ModelToINetworkConverter.cpp b/ModelToINetworkConverter.cpp index 24fb4903..0d6ddc34 100644 --- a/ModelToINetworkConverter.cpp +++ b/ModelToINetworkConverter.cpp @@ -59,8 +59,17 @@ void ModelToINetworkConverter<HalPolicy>::Convert() totalPoolSize += pool.size(); } + using NetworkOptions = std::vector<armnn::BackendOptions>; + NetworkOptions networkOptions; + armnn::BackendOptions shapeInferenceMethodOption("ShapeInferenceMethod", + { + { "InferAndValidate", true } + }); + + networkOptions.push_back(shapeInferenceMethodOption); + // Create armnn::INetwork - m_Data.m_Network = armnn::INetwork::Create(); + m_Data.m_Network = armnn::INetwork::Create(networkOptions); // add operations to it // track which layer outputs each operand |