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

[IE Samples] Export and Import of model, fixed (#9776)

Browse Source 
* [IE Samples] Export and Import of model, fixed https://jira.devtools.intel.com/browse/CVS-76225

* Added IE_SUPPRESS_DEPRECATED

* [IE Samples] Export and Import of model, fixed 76225

* Added IE_SUPPRESS_DEPRECATED

* fixed notes

* Update samples/cpp/speech_sample/main.cpp

Co-authored-by: Mikhail Ryzhov <mikhail.ryzhov@intel.com>

* Update samples/cpp/speech_sample/main.cpp

Co-authored-by: Mikhail Ryzhov <mikhail.ryzhov@intel.com>

* fixed formatting

* changed processing of inputs

* working export

Co-authored-by: Mikhail Ryzhov <mikhail.ryzhov@intel.com>
This commit is contained in:
Maxim Gordeev 2022-01-27 14:46:02 +03:00 committed by GitHub
parent cf80eda8e0
commit 0627fe857e
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
1 changed files with 25 additions and 16 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

41
samples/cpp/speech_sample/main.cpp
View File

@ -83,17 +83,25 @@ int main(int argc, char* argv[]) {
// -------------------------------------
ov::Core core;
slog::info << "Loading model files:" << slog::endl << FLAGS_m << slog::endl;
std::shared_ptr<ov::Model> model = core.read_model(FLAGS_m);
check_number_of_inputs(model->inputs().size(), numInputFiles);
const ov::Layout tensor_layout{"NC"};
ov::preprocess::PrePostProcessor proc(model);
for (int i = 0; i < model->inputs().size(); i++) {
proc.input(i).tensor().set_element_type(ov::element::f32).set_layout(tensor_layout);
uint32_t batchSize = (FLAGS_cw_r > 0 || FLAGS_cw_l > 0) ? 1 : (uint32_t)FLAGS_bs;
std::shared_ptr<ov::Model> model;
// ------------------------------ Preprocessing ------------------------------------------------------
// the preprocessing steps can be done only for loaded network and are not applicable for the imported network
// (already compiled)
if (!FLAGS_m.empty()) {
model = core.read_model(FLAGS_m);
check_number_of_inputs(model->inputs().size(), numInputFiles);
const ov::Layout tensor_layout{"NC"};
ov::preprocess::PrePostProcessor proc(model);
for (int i = 0; i < model->inputs().size(); i++) {
proc.input(i).tensor().set_element_type(ov::element::f32).set_layout(tensor_layout);
}
for (int i = 0; i < model->outputs().size(); i++) {
proc.output(i).tensor().set_element_type(ov::element::f32);
}
model = proc.build();
ov::set_batch(model, batchSize);
}
for (int i = 0; i < model->outputs().size(); i++) {
proc.output(i).tensor().set_element_type(ov::element::f32);
}
model = proc.build();
// ------------------------------ Get Available Devices ------------------------------------------------------
auto isFeature = [&](const std::string xFeature) {
return FLAGS_d.find(xFeature) != std::string::npos;
@ -101,8 +109,6 @@ int main(int argc, char* argv[]) {
bool useGna = isFeature("GNA");
bool useHetero = isFeature("HETERO");
std::string deviceStr = useHetero && useGna ? "HETERO:GNA,CPU" : FLAGS_d.substr(0, (FLAGS_d.find("_")));
uint32_t batchSize = (FLAGS_cw_r > 0 || FLAGS_cw_l > 0) ? 1 : (uint32_t)FLAGS_bs;
ov::set_batch(model, batchSize);
// -----------------------------------------------------------------------------------------------------
// --------------------------- Set parameters and scale factors -------------------------------------
/** Setting parameter for per layer metrics **/
@ -221,14 +227,17 @@ int main(int argc, char* argv[]) {
executableNet = core.compile_model(model, deviceStr, genericPluginConfig);
} else {
slog::info << "Importing model to the device" << slog::endl;
std::istringstream streamrq(FLAGS_rg);
std::ifstream streamrq(FLAGS_rg, std::ios_base::binary | std::ios_base::in);
if (!streamrq.is_open()) {
throw std::runtime_error("Cannot open model file " + FLAGS_rg);
}
executableNet = core.import_model(streamrq, deviceStr, genericPluginConfig);
}
// --------------------------- Exporting gna model using InferenceEngine AOT API---------------------
if (!FLAGS_wg.empty()) {
slog::info << "Writing GNA Model to file " << FLAGS_wg << slog::endl;
t0 = Time::now();
std::ostringstream streamwq(FLAGS_wg);
std::ofstream streamwq(FLAGS_wg, std::ios_base::binary | std::ios::out);
executableNet.export_model(streamwq);
ms exportTime = std::chrono::duration_cast<ms>(Time::now() - t0);
slog::info << "Exporting time " << exportTime.count() << " ms" << slog::endl;
@ -455,10 +464,10 @@ int main(int argc, char* argv[]) {
}
// -----------------------------------------------------------------------------------------------------
int index = static_cast<int>(frameIndex) - (FLAGS_cw_l + FLAGS_cw_r);
for (int i = 0; i < model->inputs().size(); i++) {
for (int i = 0; i < executableNet.inputs().size(); i++) {
inferRequest.inferRequest.set_input_tensor(
i,
ov::Tensor(ov::element::f32, model->inputs()[i].get_shape(), inputFrame[0]));
ov::Tensor(ov::element::f32, executableNet.inputs()[i].get_shape(), inputFrame[0]));
}
/* Starting inference in asynchronous mode*/
inferRequest.inferRequest.start_async();