TensorRT教程7：使用 C++ parser API解析模型（重点）

使用 C++ parser API解析模型

1、parser解析caffe模型

//step1:创建logger：日志记录器
class Logger : public ILogger           
 {
     void log(Severity severity, const char* msg) override
     {
         // suppress info-level messages
         if (severity != Severity::kINFO)
             std::cout << msg << std::endl;
     }
 } gLogger;

//step2:创建builder
IBuilder* builder = createInferBuilder(gLogger);
//step3:创建network
INetworkDefinition* network = builder->createNetworkV2(0U);
//step4:创建parser
ICaffeParser* parser = createCaffeParser();
//step5:使用parser，填充network对象   注意： DataType::kFLOAT（fp32）   DataType::kHALF(fp16)
const IBlobNameToTensor* blobNameToTensor = parser->parse("deploy_file",//定义网络文件
                                                          "modefile",//权重文件
                                                          *network,//解析后填充到network对象
                                                          DataType::kFLOAT);//fp32
assert(blobNameToTensor != nullptr);

//step6:标记网络输出
for (auto &s : OUTPUT_BLOB_NAMES)
    network->markOutput(*blobNameToTensor->find(s.c_str()));

//step7:创建config并设置最大batchsize和最大工作空间
IBuilderConfig* config = builder->createBuilderConfig();
config->setMaxBatchSize(maxBatchSize);//设置最大batchsize
config->setMaxWorkspaceSize(1 << 30);//2^30 ,这里是1G
//step8:创建engine
ICudaEngine* engine = builder->buildEngineWithConfig(*network, *config);
assert(engine);

//step9:序列化保存engine到planfile
IHostMemory *serializedModel = engine->serialize();
assert(serializedModel != nullptr)
std::ofstream p("xxxxx.engine");
p.write(reinterpret_cast<const char*>(serializedModel->data()), serializedModel->size());

//step10:释放资源
serializedModel->destroy();
engine->destroy();
parser->destroy()
network->destroy();
config->destroy();
builder->destroy();

2、parser解析onnx模型

在 TensorRT 7.0 中，ONNX 解析器仅支持全维模式，这意味着您的网络定义必须使用 explicitBatch 标志集创建

//step1:创建logger：日志记录器
class Logger : public ILogger           
 {
     void log(Severity severity, const char* msg) override
     {
         // suppress info-level messages
         if (severity != Severity::kINFO)
             std::cout << msg << std::endl;
     }
 } gLogger;


//step2:创建builder
IBuilder* builder = createInferBuilder(gLogger);

//step3:创建network
const auto explicitBatch = 1U << static_cast<uint32_t>(NetworkDefinitionCreationFlag::kEXPLICIT_BATCH);  
INetworkDefinition* network = builder->createNetworkV2(explicitBatch);

//step4:创建parser
nvonnxparser::IParser* parser = nvonnxparser::createParser(*network, gLogger);

//step5:使用parser解析模型填充network
const char* onnx_filename="./model.onnx"
parser->parseFromFile(onnx_filename, ILogger::Severity::kWARNING);
for (int i = 0; i < parser.getNbErrors(); ++i)
{
    std::cout << parser->getError(i)->desc() << std::endl;
}

//step6:标记网络输出
for (auto &s : OUTPUT_BLOB_NAMES)
    network->markOutput(*blobNameToTensor->find(s.c_str()));

//step7:创建config并设置最大batchsize和最大工作空间
IBuilderConfig* config = builder->createBuilderConfig();
config->setMaxBatchSize(maxBatchSize);//设置最大batchsize
config->setMaxWorkspaceSize(1 << 30);//2^30 ,这里是1G

//step8:创建engine
ICudaEngine* engine = builder->buildEngineWithConfig(*network, *config);
assert(engine);

//step9:序列化保存engine到planfile
IHostMemory *serializedModel = engine->serialize();
assert(serializedModel != nullptr)
std::ofstream p("xxxxx.engine");
p.write(reinterpret_cast<const char*>(serializedModel->data()), serializedModel->size());

//step10:释放资源
serializedModel->destroy();
engine->destroy();
parser->destroy()
network->destroy();
config->destroy();
builder->destroy();

3、parser解析UFF模型

UFF格式的模型是tensorflow导出的中间模型，UFF 不支持explicitBatch 或动态形状网络。

//step1:创建logger：日志记录器
class Logger : public ILogger           
 {
     void log(Severity severity, const char* msg) override
     {
         // suppress info-level messages
         if (severity != Severity::kINFO)
             std::cout << msg << std::endl;
     }
 } gLogger;

//step2:创建builder
IBuilder* builder = createInferBuilder(gLogger);
//step3:创建network
INetworkDefinition* network = builder->createNetworkV2(0U);
//step4:创建parser
IUFFParser* parser = createUffParser();

//step5:使用parser，填充network对象   注意： DataType::kFLOAT（fp32）   DataType::kHALF(fp16)
const IBlobNameToTensor* blobNameToTensor = parser->parse(uffFile,//模型文件
                                                          *network,//解析后填充到network对象
                                                          DataType::kFLOAT);//数据类型
assert(blobNameToTensor != nullptr);

//step6:注册网络输入输出
parser->registerInput("Input_0", DimsCHW(1, 28, 28), UffInputOrder::kNCHW);
parser->registerOutput("Binary_3");

//step7:创建config并设置最大batchsize和最大工作空间
IBuilderConfig* config = builder->createBuilderConfig();
config->setMaxBatchSize(maxBatchSize);//设置最大batchsize
config->setMaxWorkspaceSize(1 << 30);//2^30 ,这里是1G
//step8:创建engine
ICudaEngine* engine = builder->buildEngineWithConfig(*network, *config);
assert(engine);

//step9:序列化保存engine到planfile
IHostMemory *serializedModel = engine->serialize();
assert(serializedModel != nullptr)
std::ofstream p("xxxxx.engine");
p.write(reinterpret_cast<const char*>(serializedModel->data()), serializedModel->size());

//step10:释放资源
serializedModel->destroy();
engine->destroy();
parser->destroy()
network->destroy();
config->destroy();
builder->destroy();