opencv/modules/dnn/perf/perf_net.cpp
2017-09-25 15:32:37 +03:00

150 lines
4.6 KiB
C++

// This file is part of OpenCV project.
// It is subject to the license terms in the LICENSE file found in the top-level directory
// of this distribution and at http://opencv.org/license.html.
//
// Copyright (C) 2017, Intel Corporation, all rights reserved.
// Third party copyrights are property of their respective owners.
#include "perf_precomp.hpp"
#include "opencv2/core/ocl.hpp"
#include "opencv2/dnn/shape_utils.hpp"
namespace
{
#ifdef HAVE_HALIDE
#define TEST_DNN_BACKEND DNN_BACKEND_DEFAULT, DNN_BACKEND_HALIDE
#else
#define TEST_DNN_BACKEND DNN_BACKEND_DEFAULT
#endif
#define TEST_DNN_TARGET DNN_TARGET_CPU, DNN_TARGET_OPENCL
CV_ENUM(DNNBackend, DNN_BACKEND_DEFAULT, DNN_BACKEND_HALIDE)
CV_ENUM(DNNTarget, DNN_TARGET_CPU, DNN_TARGET_OPENCL)
class DNNTestNetwork : public ::perf::TestBaseWithParam< tuple<DNNBackend, DNNTarget> >
{
public:
dnn::Backend backend;
dnn::Target target;
dnn::Net net;
void processNet(std::string weights, std::string proto, std::string halide_scheduler,
int inWidth, int inHeight, const std::string& outputLayer,
const std::string& framework)
{
backend = (dnn::Backend)(int)get<0>(GetParam());
target = (dnn::Target)(int)get<1>(GetParam());
if (backend == DNN_BACKEND_DEFAULT && target == DNN_TARGET_OPENCL)
{
#if 0 //defined(HAVE_OPENCL)
if (!cv::ocl::useOpenCL())
#endif
{
throw ::SkipTestException("OpenCL is not available/disabled in OpenCV");
}
}
Mat input(inHeight, inWidth, CV_32FC3);
randu(input, 0.0f, 1.0f);
weights = findDataFile(weights, false);
if (!proto.empty())
proto = findDataFile(proto, false);
if (!halide_scheduler.empty() && backend == DNN_BACKEND_HALIDE)
halide_scheduler = findDataFile(std::string("dnn/halide_scheduler_") + (target == DNN_TARGET_OPENCL ? "opencl_" : "") + halide_scheduler, true);
if (framework == "caffe")
{
net = cv::dnn::readNetFromCaffe(proto, weights);
}
else if (framework == "torch")
{
net = cv::dnn::readNetFromTorch(weights);
}
else if (framework == "tensorflow")
{
net = cv::dnn::readNetFromTensorflow(weights);
}
else
CV_Error(Error::StsNotImplemented, "Unknown framework " + framework);
net.setInput(blobFromImage(input, 1.0, Size(), Scalar(), false));
net.setPreferableBackend(backend);
net.setPreferableTarget(target);
if (backend == DNN_BACKEND_HALIDE)
{
net.setHalideScheduler(halide_scheduler);
}
MatShape netInputShape = shape(1, 3, inHeight, inWidth);
size_t weightsMemory = 0, blobsMemory = 0;
net.getMemoryConsumption(netInputShape, weightsMemory, blobsMemory);
int64 flops = net.getFLOPS(netInputShape);
net.forward(outputLayer); // warmup
std::cout << "Memory consumption:" << std::endl;
std::cout << " Weights(parameters): " << divUp(weightsMemory, 1u<<20) << " Mb" << std::endl;
std::cout << " Blobs: " << divUp(blobsMemory, 1u<<20) << " Mb" << std::endl;
std::cout << "Calculation complexity: " << flops * 1e-9 << " GFlops" << std::endl;
PERF_SAMPLE_BEGIN()
net.forward();
PERF_SAMPLE_END()
SANITY_CHECK_NOTHING();
}
};
PERF_TEST_P_(DNNTestNetwork, AlexNet)
{
processNet("dnn/bvlc_alexnet.caffemodel", "dnn/bvlc_alexnet.prototxt",
"alexnet.yml", 227, 227, "prob", "caffe");
}
PERF_TEST_P_(DNNTestNetwork, GoogLeNet)
{
processNet("dnn/bvlc_googlenet.caffemodel", "dnn/bvlc_googlenet.prototxt",
"", 224, 224, "prob", "caffe");
}
PERF_TEST_P_(DNNTestNetwork, ResNet50)
{
processNet("dnn/ResNet-50-model.caffemodel", "dnn/ResNet-50-deploy.prototxt",
"resnet_50.yml", 224, 224, "prob", "caffe");
}
PERF_TEST_P_(DNNTestNetwork, SqueezeNet_v1_1)
{
processNet("dnn/squeezenet_v1.1.caffemodel", "dnn/squeezenet_v1.1.prototxt",
"squeezenet_v1_1.yml", 227, 227, "prob", "caffe");
}
PERF_TEST_P_(DNNTestNetwork, Inception_5h)
{
processNet("dnn/tensorflow_inception_graph.pb", "",
"inception_5h.yml",
224, 224, "softmax2", "tensorflow");
}
PERF_TEST_P_(DNNTestNetwork, ENet)
{
processNet("dnn/Enet-model-best.net", "", "enet.yml",
512, 256, "l367_Deconvolution", "torch");
}
INSTANTIATE_TEST_CASE_P(/*nothing*/, DNNTestNetwork,
testing::Combine(
::testing::Values(TEST_DNN_BACKEND),
DNNTarget::all()
)
);
} // namespace