An error occurred while loading the file. Please try again.
-
Remi Cresson authored7fb41950
/*========================================================================= Copyright (c) 2018-2019 IRSTEA Copyright (c) 2020-2021 INRAE This software is distributed WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the above copyright notices for more information.=========================================================================*/#ifndef otbTensorflowMultisourceModelBase_txx#define otbTensorflowMultisourceModelBase_txx#include "otbTensorflowMultisourceModelBase.h"namespace otb{template <class TInputImage, class TOutputImage>TensorflowMultisourceModelBase<TInputImage, TOutputImage>::TensorflowMultisourceModelBase(){ Superclass::SetCoordinateTolerance(itk::NumericTraits<double>::max()); Superclass::SetDirectionTolerance(itk::NumericTraits<double>::max()); m_SavedModel = NULL;}template <class TInputImage, class TOutputImage>tensorflow::SignatureDefTensorflowMultisourceModelBase<TInputImage, TOutputImage>::GetSignatureDef(){ auto signatures = this->GetSavedModel()->GetSignatures(); tensorflow::SignatureDef signature_def; if (signatures.size() == 0) { itkExceptionMacro("There are no available signatures for this tag-set. \n" << "Please check which tag-set to use by running " << "`saved_model_cli show --dir your_model_dir --all`"); } // If serving_default key exists (which is the default for TF saved model), choose it as signature // Else, choose the first one if (signatures.contains(tensorflow::kDefaultServingSignatureDefKey)) { signature_def = signatures.at(tensorflow::kDefaultServingSignatureDefKey); } else { signature_def = signatures.begin()->second; } return signature_def;}template <class TInputImage, class TOutputImage>voidTensorflowMultisourceModelBase<TInputImage, TOutputImage>::PushBackInputTensorBundle( std::string placeholder, SizeType receptiveField, ImagePointerType image, bool useNodata, InternalPixelType nodataValue){ Superclass::PushBackInput(image); m_InputReceptiveFields.push_back(receptiveField); m_InputPlaceholders.push_back(placeholder); m_InputUseNodata.push_back(useNodata); m_InputNodataValues.push_back(nodataValue);}7172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140
template <class TInputImage, class TOutputImage>
std::stringstream
TensorflowMultisourceModelBase<TInputImage, TOutputImage>::GenerateDebugReport(DictType & inputs)
{
// Create a debug report
std::stringstream debugReport;
// Describe the output buffered region
ImagePointerType outputPtr = this->GetOutput();
const RegionType outputReqRegion = outputPtr->GetRequestedRegion();
debugReport << "Output image buffered region: " << outputReqRegion << "\n";
// Describe inputs
for (unsigned int i = 0; i < this->GetNumberOfInputs(); i++)
{
const ImagePointerType inputPtr = const_cast<TInputImage *>(this->GetInput(i));
const RegionType reqRegion = inputPtr->GetRequestedRegion();
debugReport << "Input #" << i << ":\n";
debugReport << "Requested region: " << reqRegion << "\n";
debugReport << "Tensor \"" << inputs[i].first << "\": " << tf::PrintTensorInfos(inputs[i].second) << "\n";
}
// Show user placeholders
debugReport << "User placeholders:\n";
for (auto & dict : this->GetUserPlaceholders())
{
debugReport << "Tensor \"" << dict.first << "\": " << tf::PrintTensorInfos(dict.second) << "\n" << std::endl;
}
return debugReport;
}
template <class TInputImage, class TOutputImage>
void
TensorflowMultisourceModelBase<TInputImage, TOutputImage>::RunSession(DictType & inputs, TensorListType & outputs, bool & nodata)
{
// Run the TF session here
// The session will initialize the outputs
// `inputs` corresponds to a mapping {name, tensor}, with the name being specified by the user when calling
// TensorFlowModelServe we must adapt it to `inputs_new`, that corresponds to a mapping {layerName, tensor}, with the
// layerName being from the model
DictType inputs_new;
// Add the user's placeholders
std::size_t k = 0;
for (auto & dict : this->GetUserPlaceholders())
{
inputs_new.emplace_back(m_InputConstants[k], dict.second);
k++;
}
// Add input tensors
// During this step we also check for nodata values
nodata = false;
k = 0;
for (auto & dict : inputs)
{
auto inputTensor = dict.second;
inputs_new.emplace_back(m_InputLayers[k], inputTensor);
if (m_InputUseNodata[k] == true)
{
const auto nodataValue = m_InputNodataValues[k];
const tensorflow::int64 nElmT = inputTensor.NumElements();
tensorflow::int64 ndCount = 0;
auto array = inputTensor.flat<InternalPixelType>();
for (tensorflow::int64 i = 0 ; i < nElmT ; i++)
if (array(i) == nodataValue)
141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210
ndCount++;
if (ndCount == nElmT)
{
nodata = true;
return;
}
}
k += 1;
}
// Run the session, evaluating our output tensors from the graph
auto status = this->GetSavedModel()->session.get()->Run(inputs_new, m_OutputLayers, m_TargetNodesNames, &outputs);
if (!status.ok())
{
// Create a debug report
std::stringstream debugReport = GenerateDebugReport(inputs);
// Throw an exception with the report
itkExceptionMacro("Can't run the tensorflow session !\n"
<< "Tensorflow error message:\n"
<< status.ToString()
<< "\n"
"OTB Filter debug message:\n"
<< debugReport.str());
}
}
template <class TInputImage, class TOutputImage>
void
TensorflowMultisourceModelBase<TInputImage, TOutputImage>::RunSession(DictType & inputs, TensorListType & outputs)
{
bool nodata;
this->RunSession(inputs, outputs, nodata);
}
template <class TInputImage, class TOutputImage>
void
TensorflowMultisourceModelBase<TInputImage, TOutputImage>::GenerateOutputInformation()
{
// Check that the number of the following is the same
// - input placeholders names
// - input receptive fields
// - input images
const unsigned int nbInputs = this->GetNumberOfInputs();
if (nbInputs != m_InputReceptiveFields.size() || nbInputs != m_InputPlaceholders.size())
{
itkExceptionMacro("Number of input images is "
<< nbInputs << " but the number of input patches size is " << m_InputReceptiveFields.size()
<< " and the number of input tensors names is " << m_InputPlaceholders.size());
}
// Check that no-data values size is consistent with the inputs
// If no value is specified, set a vector of the same size as the inputs
if (m_InputNodataValues.size() == 0 && m_InputUseNodata.size() == 0)
{
m_InputUseNodata = BoolListType(nbInputs, false);
m_InputNodataValues = ValueListType(nbInputs, 0.0);
}
if (nbInputs != m_InputNodataValues.size() || nbInputs != m_InputUseNodata.size())
{
itkExceptionMacro("Number of input images is " << nbInputs << " but the number of no-data values is not consistent");
}
//////////////////////////////////////////////////////////////////////////////////////////
// Get tensors information
//////////////////////////////////////////////////////////////////////////////////////////
// Set all subelement of the model
211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248
auto signaturedef = this->GetSignatureDef();
// Given the inputs/outputs names that the user specified, get the names of the inputs/outputs contained in the model
// and other infos (shapes, dtypes)
// For example, for output names specified by the user m_OutputTensors = ['s2t', 's2t_pad'],
// this will return m_OutputLayers = ['PartitionedCall:0', 'PartitionedCall:1']
// In case the user hasn't named the output, i.e. m_OutputTensors = [''],
// this will return the first output m_OutputLayers = ['PartitionedCall:0']
StringList constantsNames;
std::transform(m_UserPlaceholders.begin(),
m_UserPlaceholders.end(),
std::back_inserter(constantsNames),
[](const DictElementType & p) { return p.first; });
if (m_UserPlaceholders.size() > 0)
{
// Avoid the unnecessary warning when no placeholder is fed
tf::GetTensorAttributes(signaturedef.inputs(),
constantsNames,
m_InputConstants,
m_InputConstantsShapes,
m_InputConstantsDataTypes);
}
tf::GetTensorAttributes(signaturedef.inputs(),
m_InputPlaceholders,
m_InputLayers,
m_InputTensorsShapes,
m_InputTensorsDataTypes,
constantsNames);
tf::GetTensorAttributes(
signaturedef.outputs(), m_OutputTensors, m_OutputLayers, m_OutputTensorsShapes, m_OutputTensorsDataTypes);
}
} // end namespace otb
#endif