+

#include "../InferenceTest.hpp"
+#include "../ImagePreprocessor.hpp"
+#include "armnnOnnxParser/IOnnxParser.hpp"
+

+

+ + + + +

+Functions
int	main (int argc, char *argv[])

+

Function Documentation

+ +

◆ main()

+ +

+

+ + + + + + + + + + + + + + + + + + +

int main	(	int	argc,
		char *	argv[]
	)

+

+ +

Definition at line 9 of file OnnxMobileNet-Armnn.cpp.

+

 {
     int retVal = EXIT_FAILURE;
     try
     {
         // Coverity fix: The following code may throw an exception of type std::length_error.
         std::vector<ImageSet> imageSet =
         {
             {"Dog.jpg", 208},
             {"Cat.jpg", 281},
             {"shark.jpg", 2},
         };
 
         armnn::TensorShape inputTensorShape({ 1, 3, 224, 224 });
 
         using DataType = float;
         using DatabaseType = ImagePreprocessor<float>;
         using ParserType = armnnOnnxParser::IOnnxParser;
         using ModelType = InferenceModel<ParserType, DataType>;
 
         // Coverity fix: ClassifierInferenceTestMain() may throw uncaught exceptions.
         retVal = armnn::test::ClassifierInferenceTestMain<DatabaseType, ParserType>(
                      argc, argv,
                      "mobilenetv2-1.0.onnx", // model name
                      true,                           // model is binary
                      "data", "mobilenetv20_output_flatten0_reshape0", // input and output tensor names
                      { 0, 1, 2 },                    // test images to test with as above
                      [&imageSet](const char* dataDir, const ModelType&) {
                          // This creates create a 1, 3, 224, 224 normalized input with mean and stddev to pass to Armnn
                          return DatabaseType(
                              dataDir,
                              224,
                              224,
                              imageSet,
                              255.0,                           // scale
                              {{0.485f, 0.456f, 0.406f}},      // mean
                              {{0.229f, 0.224f, 0.225f}},      // stddev
                              DatabaseType::DataFormat::NCHW); // format
                      },
                      &inputTensorShape);
     }
     catch (const std::exception& e)
     {
         // Coverity fix: BOOST_LOG_TRIVIAL (typically used to report errors) may throw an
         // exception of type std::length_error.
         // Using stderr instead in this context as there is no point in nesting try-catch blocks here.
         std::cerr << "WARNING: OnnxMobileNet-Armnn: An error has occurred when running "
                      "the classifier inference tests: " << e.what() << std::endl;
     }
     return retVal;
 }

+
+
+
+
+

+

+Functions

Function Documentation

◆ main()