opencv/modules/dnn/test/test_halide_layers.cpp

// 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) 2017-2019, Intel Corporation, all rights reserved.
// Third party copyrights are property of their respective owners.

// This tests doesn't require any external data. They just compare outputs of
// layers using different computation backends. Input and parameters are random.

#include "test_precomp.hpp"

namespace opencv_test { namespace {

using namespace cv;
using namespace cv::dnn;
using namespace testing;

static void test(Mat& input, Net& net, Backend backendId, Target targetId, bool skipCheck = false, bool randInput = true, double l1 = 0.0, double lInf = 0.0)
{
    DNNTestLayer::checkBackend(backendId, targetId);
    if (randInput)
        randu(input, -1.0f, 1.0f);

    net.setInput(input);
    net.setPreferableBackend(DNN_BACKEND_OPENCV);
    Mat outputDefault = net.forward().clone();

    net.setPreferableBackend(backendId);
    net.setPreferableTarget(targetId);
    Mat outputHalide = net.forward().clone();

    if (skipCheck)
        return;

    double default_l1, default_lInf;
    DNNTestLayer::getDefaultThresholds(backendId, targetId, &default_l1, &default_lInf);
    if (l1 == 0.0)
        l1 = default_l1;
    if (lInf == 0.0)
        lInf = default_lInf;
#if 0
    std::cout << "l1=" << l1 << "  lInf=" << lInf << std::endl;
    std::cout << outputDefault.reshape(1, outputDefault.total()).t() << std::endl;
    std::cout << outputHalide.reshape(1, outputDefault.total()).t() << std::endl;
#endif
    normAssert(outputDefault, outputHalide, "", l1, lInf);
}

static void test(LayerParams& params, Mat& input, Backend backendId, Target targetId, bool skipCheck = false, double l1 = 0.0, double lInf = 0.0)
{
    Net net;
    net.addLayerToPrev(params.name, params.type, params);
    test(input, net, backendId, targetId, skipCheck, true, l1, lInf);
}

static inline testing::internal::ParamGenerator<tuple<Backend, Target> > dnnBackendsAndTargetsWithHalide()
{
    return dnnBackendsAndTargets(true, true, false); // OpenCV/CPU is used as reference
}

class Test_Halide_layers : public DNNTestLayer {};

////////////////////////////////////////////////////////////////////////////////
// Padding
////////////////////////////////////////////////////////////////////////////////
TEST_P(Test_Halide_layers, Padding)
{
    static const int kNumRuns = 10;
    std::vector<int> paddings(8);
    cv::RNG& rng = cv::theRNG();
    for (int t = 0; t < kNumRuns; ++t)
    {
        for (int i = 0; i < paddings.size(); ++i)
            paddings[i] = rng(5);

        LayerParams lp;
        lp.set("paddings", DictValue::arrayInt<int*>(&paddings[0], paddings.size()));
        lp.type = "Padding";
        lp.name = "testLayer";

        int sz[] = {1 + (int)rng(10), 1 + (int)rng(10), 1 + (int)rng(10), 1 + (int)rng(10)};
        Mat input(4, &sz[0], CV_32F);
        test(lp, input, backend, target);
    }
}

////////////////////////////////////////////////////////////////////////////////
// Convolution
////////////////////////////////////////////////////////////////////////////////
typedef TestWithParam<tuple<Vec3i, Size, Size, Size, Size, Size, bool, tuple<Backend, Target> > > Convolution;
TEST_P(Convolution, Accuracy)
{
    int inChannels = get<0>(GetParam())[0];
    int outChannels = get<0>(GetParam())[1];
    int group = get<0>(GetParam())[2];
    Size inSize = get<1>(GetParam());
    Size kernel = get<2>(GetParam());
    Size stride = get<3>(GetParam());
    Size pad = get<4>(GetParam());
    Size dilation = get<5>(GetParam());
    bool hasBias = get<6>(GetParam());
    Backend backendId = get<0>(get<7>(GetParam()));
    Target targetId = get<1>(get<7>(GetParam()));

    bool skipCheck = false;

    int sz[] = {outChannels, inChannels / group, kernel.height, kernel.width};
    Mat weights(4, &sz[0], CV_32F);
    randu(weights, -1.0f, 1.0f);

    LayerParams lp;
    lp.set("kernel_w", kernel.width);
    lp.set("kernel_h", kernel.height);
    lp.set("pad_w", pad.width);
    lp.set("pad_h", pad.height);
    lp.set("stride_w", stride.width);
    lp.set("stride_h", stride.height);
    lp.set("dilation_w", dilation.width);
    lp.set("dilation_h", dilation.height);
    lp.set("num_output", outChannels);
    lp.set("group", group);
    lp.set("bias_term", hasBias);
    lp.type = "Convolution";
    lp.name = "testLayer";
    lp.blobs.push_back(weights);
    if (hasBias)
    {
        Mat bias(1, outChannels, CV_32F);
        randu(bias, -1.0f, 1.0f);
        lp.blobs.push_back(bias);
    }
    int inpSz[] = {1, inChannels, inSize.height, inSize.width};
    Mat input(4, &inpSz[0], CV_32F);
    test(lp, input, backendId, targetId, skipCheck);
    if (skipCheck)
        throw SkipTestException("Skip checks in unstable test");
}

INSTANTIATE_TEST_CASE_P(Layer_Test_Halide, Convolution, Combine(
/*in channels, out channels, group*/
             Values(Vec3i(6, 4, 1), Vec3i(6, 9, 1),
                    Vec3i(6, 4, 2), Vec3i(6, 9, 3)),
/*in size*/  Values(Size(5, 6)),
/*kernel*/   Values(Size(3, 1), Size(1, 3)),
/*stride*/   Values(Size(1, 1), Size(2, 2)),
/*pad*/      Values(Size(1, 0), Size(0, 1)),
/*dilation*/ Values(Size(1, 1), Size(2, 2)),
/*has bias*/ Bool(),
             dnnBackendsAndTargetsWithHalide()
));

////////////////////////////////////////////////////////////////////////////////
// Deconvolution
////////////////////////////////////////////////////////////////////////////////
typedef TestWithParam<tuple<Vec3i, Size, Size, Size, Size, Vec4i, bool, tuple<Backend, Target> > > Deconvolution;
TEST_P(Deconvolution, Accuracy)
{
    int inChannels = get<0>(GetParam())[0];
    int outChannels = get<0>(GetParam())[1];
    int group = get<0>(GetParam())[2];
    Size inSize = get<1>(GetParam());
    Size kernel = get<2>(GetParam());
    Size pad = get<3>(GetParam());
    Size dilation = get<4>(GetParam());
    Size stride = Size(get<5>(GetParam())[0], get<5>(GetParam())[1]);
    Size adjPad = Size(get<5>(GetParam())[2], get<5>(GetParam())[3]);
    bool hasBias = get<6>(GetParam());
    Backend backendId = get<0>(get<7>(GetParam()));
    Target targetId = get<1>(get<7>(GetParam()));

#if defined(INF_ENGINE_RELEASE) && INF_ENGINE_VER_MAJOR_GE(2019010000)
    if (backendId == DNN_BACKEND_INFERENCE_ENGINE_NN_BUILDER_2019 && targetId == DNN_TARGET_MYRIAD
            && getInferenceEngineVPUType() == CV_DNN_INFERENCE_ENGINE_VPU_TYPE_MYRIAD_X
            && inChannels == 6 && outChannels == 4 && group == 1
            && kernel == Size(1, 3) && pad == Size(1, 0)
            && stride == Size(1, 1) && dilation == Size(1, 1))
        applyTestTag(CV_TEST_TAG_DNN_SKIP_IE_MYRIAD_X);
#endif

    if (targetId == DNN_TARGET_CUDA_FP16)
        applyTestTag(CV_TEST_TAG_DNN_SKIP_CUDA_FP16);

    int sz[] = {inChannels, outChannels / group, kernel.height, kernel.width};
    Mat weights(4, &sz[0], CV_32F);
    randu(weights, -1.0f, 1.0f);

    LayerParams lp;
    lp.set("kernel_w", kernel.width);
    lp.set("kernel_h", kernel.height);
    lp.set("pad_w", pad.width);
    lp.set("pad_h", pad.height);
    lp.set("stride_w", stride.width);
    lp.set("stride_h", stride.height);
    lp.set("dilation_w", dilation.width);
    lp.set("dilation_h", dilation.height);
    lp.set("adj_w", adjPad.width);
    lp.set("adj_h", adjPad.height);
    lp.set("num_output", outChannels);
    lp.set("group", group);
    lp.set("bias_term", hasBias);
    lp.type = "Deconvolution";
    lp.name = "testLayer";
    lp.blobs.push_back(weights);
    if (hasBias)
    {
        Mat bias(1, outChannels, CV_32F);
        randu(bias, -1.0f, 1.0f);
        lp.blobs.push_back(bias);
    }
    int inpSz[] = {1, inChannels, inSize.height, inSize.width};
    Mat input(4, &inpSz[0], CV_32F);
    test(lp, input, backendId, targetId);
}

INSTANTIATE_TEST_CASE_P(Layer_Test_Halide, Deconvolution, Combine(
/*in channels, out channels, group*/
             Values(Vec3i(6, 4, 1), Vec3i(6, 9, 3)),
/*in size*/  Values(Size(5, 6)),
/*kernel*/   Values(Size(3, 1), Size(1, 3)),
/*pad*/      Values(Size(1, 0), Size(0, 1)),
/*dilation*/ Values(Size(1, 1)),
/*stride, adj. pad*/ Values(Vec4i(1,1, 0,0), Vec4i(2,2, 1,0), Vec4i(1,2, 0,1)),
/*has bias*/ Bool(),
             dnnBackendsAndTargetsWithHalide()
));

////////////////////////////////////////////////////////////////////////////////
// LRN
////////////////////////////////////////////////////////////////////////////////
typedef TestWithParam<tuple<Vec3i, int, Vec3f, bool, std::string, tuple<Backend, Target> > > LRN;
TEST_P(LRN, Accuracy)
{
    int inChannels = get<0>(GetParam())[0];
    Size inSize = Size(get<0>(GetParam())[1], get<0>(GetParam())[2]);
    int localSize = get<1>(GetParam());
    float alpha = get<2>(GetParam())[0];
    float beta = get<2>(GetParam())[1];
    float bias = get<2>(GetParam())[2];
    bool normBySize = get<3>(GetParam());
    std::string nrmType = get<4>(GetParam());
    Backend backendId = get<0>(get<5>(GetParam()));
    Target targetId = get<1>(get<5>(GetParam()));

    if ((inSize.width == 5 || inSize.height == 5) && targetId == DNN_TARGET_MYRIAD &&
        nrmType == "ACROSS_CHANNELS")
        applyTestTag(CV_TEST_TAG_DNN_SKIP_IE_MYRIAD);

    LayerParams lp;
    lp.set("norm_region", nrmType);
    lp.set("local_size", localSize);
    lp.set("alpha", alpha);
    lp.set("beta", beta);
    lp.set("bias", bias);
    lp.set("norm_by_size", normBySize);
    lp.type = "LRN";
    lp.name = "testLayer";

    int sz[] = {1, inChannels, inSize.height, inSize.width};
    Mat input(4, &sz[0], CV_32F);

    double l1 = 0.0, lInf = 0.0;
    // The OpenCL kernels use the native_ math functions which have
    // implementation defined accuracy, so we use relaxed thresholds. See
    // https://github.com/opencv/opencv/issues/9821 for more details.
    if (targetId == DNN_TARGET_OPENCL)
    {
        l1 = 0.01;
        lInf = 0.01;
    }
    test(lp, input, backendId, targetId, false, l1, lInf);
}

INSTANTIATE_TEST_CASE_P(Layer_Test_Halide, LRN, Combine(
/*input ch,w,h*/ Values(Vec3i(6, 5, 8), Vec3i(7, 11, 6)),
/*local size*/   Values(3, 5),
                 Values(Vec3f(0.9f, 1.0f, 1.1f), Vec3f(0.9f, 1.1f, 1.0f),
/*alpha, beta, bias*/   Vec3f(1.0f, 0.9f, 1.1f), Vec3f(1.0f, 1.1f, 0.9f),
                        Vec3f(1.1f, 0.9f, 1.0f), Vec3f(1.1f, 1.0f, 0.9f)),
/*norm_by_size*/ Bool(),
/*norm_type*/    Values("ACROSS_CHANNELS", "WITHIN_CHANNEL"),
                 dnnBackendsAndTargetsWithHalide()
));

////////////////////////////////////////////////////////////////////////////////
// Average pooling
////////////////////////////////////////////////////////////////////////////////
typedef TestWithParam<tuple<int, Size, Size, Size, tuple<Backend, Target> > > AvePooling;
TEST_P(AvePooling, Accuracy)
{
    int inChannels = get<0>(GetParam());
    Size outSize = get<1>(GetParam());;  // Input size will be computed from parameters.
    Size kernel = get<2>(GetParam());
    Size stride = get<3>(GetParam());
    Backend backendId = get<0>(get<4>(GetParam()));
    Target targetId = get<1>(get<4>(GetParam()));

#if defined(INF_ENGINE_RELEASE)
    if (backendId == DNN_BACKEND_INFERENCE_ENGINE_NN_BUILDER_2019 && targetId == DNN_TARGET_MYRIAD
            && getInferenceEngineVPUType() == CV_DNN_INFERENCE_ENGINE_VPU_TYPE_MYRIAD_X
            && kernel == Size(1, 1) && (stride == Size(1, 1) || stride == Size(2, 2)))
        applyTestTag(CV_TEST_TAG_DNN_SKIP_IE_MYRIAD_X);
#endif

    const int inWidth = (outSize.width - 1) * stride.width + kernel.width;
    const int inHeight = (outSize.height - 1) * stride.height + kernel.height;

    LayerParams lp;
    lp.set("pool", "ave");
    lp.set("kernel_w", kernel.width);
    lp.set("kernel_h", kernel.height);
    lp.set("stride_w", stride.width);
    lp.set("stride_h", stride.height);
    lp.type = "Pooling";
    lp.name = "testLayer";

    int sz[] = {1, inChannels, inHeight, inWidth};
    Mat input(4, &sz[0], CV_32F);
    test(lp, input, backendId, targetId);
}

INSTANTIATE_TEST_CASE_P(Layer_Test_Halide, AvePooling, Combine(
/*in channels*/ Values(3, 4),
/*out size*/    Values(Size(1, 1), Size(2, 2), Size(3, 2), Size(4, 7)),
/*kernel*/      Values(Size(1, 1), Size(2, 2), Size(3, 3), Size(3, 2)),
/*stride*/      Values(Size(1, 1), Size(2, 2), Size(3, 2)),
                dnnBackendsAndTargetsWithHalide()
));

////////////////////////////////////////////////////////////////////////////////
// Maximum pooling
////////////////////////////////////////////////////////////////////////////////
typedef TestWithParam<tuple<int, Size, Size, Size, Size, tuple<Backend, Target> > > MaxPooling;
TEST_P(MaxPooling, Accuracy)
{
    int inChannels = get<0>(GetParam());
    Size inSize = get<1>(GetParam());
    Size kernel = get<2>(GetParam());
    Size stride = get<3>(GetParam());
    Size pad = get<4>(GetParam());
    Backend backendId = get<0>(get<5>(GetParam()));
    Target targetId = get<1>(get<5>(GetParam()));

#if defined(INF_ENGINE_RELEASE) && INF_ENGINE_VER_MAJOR_LE(2018050000)
    if (backendId == DNN_BACKEND_INFERENCE_ENGINE_NN_BUILDER_2019 && targetId == DNN_TARGET_MYRIAD
            && inSize == Size(7, 6) && kernel == Size(3, 2)
            && (stride == Size(1, 1) || stride == Size(2, 2))
            && (pad == Size(0, 1) || pad == Size(1, 1))
    )
        applyTestTag(CV_TEST_TAG_DNN_SKIP_IE_MYRIAD, CV_TEST_TAG_DNN_SKIP_IE_VERSION);
#endif

#if defined(INF_ENGINE_RELEASE) && INF_ENGINE_VER_MAJOR_EQ(2018050000)
    if (backendId == DNN_BACKEND_INFERENCE_ENGINE_NN_BUILDER_2019 && targetId == DNN_TARGET_MYRIAD
            && (kernel == Size(2, 2) || kernel == Size(3, 2))
            && stride == Size(1, 1) && (pad == Size(0, 0) || pad == Size(0, 1))
    )
        applyTestTag(CV_TEST_TAG_DNN_SKIP_IE_MYRIAD, CV_TEST_TAG_DNN_SKIP_IE_VERSION);
#endif

#if defined(INF_ENGINE_RELEASE) && INF_ENGINE_VER_MAJOR_GE(2019010000)
    if (backendId == DNN_BACKEND_INFERENCE_ENGINE_NN_BUILDER_2019 && targetId == DNN_TARGET_MYRIAD
            && getInferenceEngineVPUType() == CV_DNN_INFERENCE_ENGINE_VPU_TYPE_MYRIAD_X
            && (stride == Size(1, 1) || stride == Size(2, 2))
            && (pad == Size(0, 1) || pad == Size(1, 1))
    )
        applyTestTag(CV_TEST_TAG_DNN_SKIP_IE_MYRIAD_X, CV_TEST_TAG_DNN_SKIP_IE_VERSION);
#endif

#if defined(INF_ENGINE_RELEASE) && INF_ENGINE_VER_MAJOR_GE(2020020000)
    if (backendId == DNN_BACKEND_INFERENCE_ENGINE_NGRAPH && targetId == DNN_TARGET_MYRIAD)
        applyTestTag(CV_TEST_TAG_DNN_SKIP_IE_MYRIAD, CV_TEST_TAG_DNN_SKIP_IE_NGRAPH, CV_TEST_TAG_DNN_SKIP_IE_VERSION);
#endif

    LayerParams lp;
    lp.set("pool", "max");
    lp.set("kernel_w", kernel.width);
    lp.set("kernel_h", kernel.height);
    lp.set("stride_w", stride.width);
    lp.set("stride_h", stride.height);
    lp.set("pad_w", pad.width);
    lp.set("pad_h", pad.height);
    lp.type = "Pooling";
    lp.name = "testLayer";

    int sz[] = {1, inChannels, inSize.height, inSize.width};
    Mat input(4, &sz[0], CV_32F);
    test(lp, input, backendId, targetId);
}

INSTANTIATE_TEST_CASE_P(Layer_Test_Halide, MaxPooling, Combine(
/*in channels*/ Values(3, 4),
/*in size*/     Values(Size(5, 5), Size(7, 6)),
/*kernel*/      Values(Size(2, 2), Size(3, 3), Size(3, 2)),
/*stride*/      Values(Size(1, 1), Size(2, 2), Size(3, 2)),
/*pad*/         Values(Size(0, 0), Size(1, 1), Size(0, 1)),
                dnnBackendsAndTargetsWithHalide()
));

////////////////////////////////////////////////////////////////////////////////
// Fully-connected
////////////////////////////////////////////////////////////////////////////////
typedef TestWithParam<tuple<int, Size, int, bool, tuple<Backend, Target> > > FullyConnected;
TEST_P(FullyConnected, Accuracy)
{
    int inChannels = get<0>(GetParam());
    Size inSize = get<1>(GetParam());
    int outChannels = get<2>(GetParam());
    bool hasBias = get<3>(GetParam());
    Backend backendId = get<0>(get<4>(GetParam()));
    Target targetId = get<1>(get<4>(GetParam()));
    if ((backendId == DNN_BACKEND_INFERENCE_ENGINE_NN_BUILDER_2019 ||
         backendId == DNN_BACKEND_INFERENCE_ENGINE_NGRAPH) && (targetId == DNN_TARGET_OPENCL_FP16 ||
       (targetId == DNN_TARGET_MYRIAD && getInferenceEngineVPUType() == CV_DNN_INFERENCE_ENGINE_VPU_TYPE_MYRIAD_X))) {
        applyTestTag(CV_TEST_TAG_DNN_SKIP_IE_OPENCL_FP16);
        applyTestTag(CV_TEST_TAG_DNN_SKIP_IE_MYRIAD_X);
    }

    Mat weights(outChannels, inChannels * inSize.height * inSize.width, CV_32F);
    randu(weights, -1.0f, 1.0f);

    Mat bias(1, outChannels, CV_32F);
    randu(bias, -1.0f, 1.0f);

    LayerParams lp;
    lp.set("num_output", outChannels);
    lp.set("bias_term", hasBias);
    lp.blobs.push_back(weights);
    lp.blobs.push_back(bias);
    lp.type = "InnerProduct";
    lp.name = "testLayer";

    int sz[] = {1, inChannels, inSize.height, inSize.width};
    Mat input(4, &sz[0], CV_32F);

    double l1 = 0.0;
    if (targetId == DNN_TARGET_CUDA_FP16)
        l1 = 0.015;
    test(lp, input, backendId, targetId, false, true, l1);
}

INSTANTIATE_TEST_CASE_P(Layer_Test_Halide, FullyConnected, Combine(
/*in channels*/  Values(3, 4),
/*in size*/      Values(Size(5, 4), Size(4, 5), Size(1, 1)),
/*out channels*/ Values(3, 4),
/*has bias*/     Bool(),
                 dnnBackendsAndTargetsWithHalide()
));

////////////////////////////////////////////////////////////////////////////////
// SoftMax
////////////////////////////////////////////////////////////////////////////////
typedef TestWithParam<tuple<int,  tuple<Backend, Target> > > SoftMax;
TEST_P(SoftMax, Accuracy)
{
    int inChannels = get<0>(GetParam());
    Backend backendId = get<0>(get<1>(GetParam()));
    Target targetId = get<1>(get<1>(GetParam()));
    LayerParams lp;
    lp.type = "Softmax";
    lp.name = "testLayer";

    int sz[] = {1, inChannels, 1, 1};
    Mat input(4, &sz[0], CV_32F);
    test(lp, input, backendId, targetId);
}

INSTANTIATE_TEST_CASE_P(Layer_Test_Halide, SoftMax, Combine(
    Values(3, 4, 5, 1024),
    dnnBackendsAndTargetsWithHalide()
));

//////////////////////////////////////////////////////////////////////////////
// Max pooling - unpooling
//////////////////////////////////////////////////////////////////////////////
TEST_P(Test_Halide_layers, MaxPoolUnpool)
{
    if (backend == DNN_BACKEND_INFERENCE_ENGINE_NN_BUILDER_2019)
        applyTestTag(CV_TEST_TAG_DNN_SKIP_IE_NN_BUILDER);
    if (backend == DNN_BACKEND_INFERENCE_ENGINE_NGRAPH)
        applyTestTag(CV_TEST_TAG_DNN_SKIP_IE_NGRAPH);

    LayerParams pool;
    pool.set("pool", "max");
    pool.set("kernel_w", 2);
    pool.set("kernel_h", 2);
    pool.set("stride_w", 2);
    pool.set("stride_h", 2);
    pool.set("pad_w", 0);
    pool.set("pad_h", 0);
    pool.type = "Pooling";
    pool.name = "testPool";

    LayerParams unpool;
    unpool.set("pool_k_w", 2);
    unpool.set("pool_k_h", 2);
    unpool.set("pool_stride_w", 2);
    unpool.set("pool_stride_h", 2);
    unpool.set("pool_pad_w", 0);
    unpool.set("pool_pad_h", 0);
    unpool.type = "MaxUnpool";
    unpool.name = "testUnpool";

    Net net;
    int poolId = net.addLayer(pool.name, pool.type, pool);
    net.connect(0, 0, poolId, 0);

    int unpoolId = net.addLayer(unpool.name, unpool.type, unpool);
    net.connect(poolId, 0, unpoolId, 0);
    net.connect(poolId, 1, unpoolId, 1);

    int sz[] = {1, 1, 4, 4};
    Mat input(4, &sz[0], CV_32F);
    test(input, net, backend, target);
}

////////////////////////////////////////////////////////////////////////////////
// AvePooling + in-place layers
////////////////////////////////////////////////////////////////////////////////
static const int kNumChannels = 3;

void testInPlaceActivation(LayerParams& lp, Backend backendId, Target targetId, double l1 = 0.0, double lInf = 0.0)
{
    EXPECT_FALSE(lp.name.empty());

    LayerParams pool;
    pool.set("pool", "ave");
    pool.set("kernel_w", 2);
    pool.set("kernel_h", 2);
    pool.set("stride_w", 2);
    pool.set("stride_h", 2);
    pool.type = "Pooling";
    pool.name = "ave_pool";

    Net net;
    int poolId = net.addLayer(pool.name, pool.type, pool);
    net.connect(0, 0, poolId, 0);
    net.addLayerToPrev(lp.name, lp.type, lp);

    int sz[] = {1, kNumChannels, 10, 10};
    Mat input(4, &sz[0], CV_32F);
    test(input, net, backendId, targetId, false, true, l1, lInf);
}

typedef TestWithParam<tuple<bool, bool, float, tuple<Backend, Target> > > BatchNorm;
TEST_P(BatchNorm, Accuracy)
{
    bool hasWeights = get<0>(GetParam());
    bool hasBias = get<1>(GetParam());
    float epsilon = get<2>(GetParam());
    Backend backendId = get<0>(get<3>(GetParam()));
    Target targetId = get<1>(get<3>(GetParam()));

    LayerParams lp;
    lp.set("has_weight", hasWeights);
    lp.set("has_bias", hasBias);
    lp.set("eps", epsilon);
    lp.type = "BatchNorm";
    lp.name = "testLayer";

    lp.blobs.reserve(4);
    for (int i = 0; i < 3; ++i)
        lp.blobs.push_back(Mat(1, kNumChannels, CV_32F));
    if (hasBias || hasWeights)
        lp.blobs.push_back(Mat(1, kNumChannels, CV_32F));

    for (int i = 0; i < lp.blobs.size(); ++i)
        randu(lp.blobs[i], 0.0f, 1.0f);

    testInPlaceActivation(lp, backendId, targetId);
}

INSTANTIATE_TEST_CASE_P(Layer_Test_Halide, BatchNorm, Combine(
/*has weights*/ Bool(),
/*has bias*/    Bool(),
/*epsilon*/     Values(1e-3f, 1e-5f),
                dnnBackendsAndTargetsWithHalide()
));

typedef TestWithParam<tuple<float, tuple<Backend, Target> > > ReLU;
TEST_P(ReLU, Accuracy)
{
    float negativeSlope = get<0>(GetParam());
    Backend backendId = get<0>(get<1>(GetParam()));
    Target targetId = get<1>(get<1>(GetParam()));

#if defined(INF_ENGINE_RELEASE) && INF_ENGINE_VER_MAJOR_GE(2019020000)
    if (backendId == DNN_BACKEND_INFERENCE_ENGINE_NN_BUILDER_2019 && targetId == DNN_TARGET_MYRIAD && negativeSlope < 0)
        applyTestTag(CV_TEST_TAG_DNN_SKIP_IE_NN_BUILDER, CV_TEST_TAG_DNN_SKIP_IE_VERSION);
#endif

    LayerParams lp;
    lp.set("negative_slope", negativeSlope);
    lp.type = "ReLU";
    lp.name = "testLayer";
    testInPlaceActivation(lp, backendId, targetId);
}

INSTANTIATE_TEST_CASE_P(Layer_Test_Halide, ReLU, Combine(
/*negative slope*/ Values(2.0f, 0.3f, -0.1f, 0.0f),
                   dnnBackendsAndTargetsWithHalide()
));

typedef TestWithParam<tuple<std::string, tuple<Backend, Target> > > NoParamActivation;
TEST_P(NoParamActivation, Accuracy)
{
    Backend backendId = get<0>(get<1>(GetParam()));
    Target targetId = get<1>(get<1>(GetParam()));

    LayerParams lp;
    lp.type = get<0>(GetParam());
    lp.name = "testLayer";
    testInPlaceActivation(lp, backendId, targetId);
}
INSTANTIATE_TEST_CASE_P(Layer_Test_Halide, NoParamActivation, Combine(
/*type*/ Values("TanH", "Sigmoid", "AbsVal", "BNLL", "Swish", "Mish"),
         dnnBackendsAndTargetsWithHalide()
));

typedef TestWithParam<tuple<Vec3f, tuple<Backend, Target> > > Power;
TEST_P(Power, Accuracy)
{
    float power = get<0>(GetParam())[0];
    float scale = get<0>(GetParam())[1];
    float shift = get<0>(GetParam())[2];
    Backend backendId = get<0>(get<1>(GetParam()));
    Target targetId = get<1>(get<1>(GetParam()));

    LayerParams lp;
    lp.set("power", power);
    lp.set("scale", scale);
    lp.set("shift", shift);
    lp.type = "Power";
    lp.name = "testLayer";
    testInPlaceActivation(lp, backendId, targetId);
}

INSTANTIATE_TEST_CASE_P(Layer_Test_Halide, Power, Combine(
/*power, scale, shift*/ Values(Vec3f(0.9f, 1.0f, 1.1f), Vec3f(0.9f, 1.1f, 1.0f),
                               Vec3f(1.0f, 0.9f, 1.1f), Vec3f(1.0f, 1.1f, 0.9f),
                               Vec3f(1.1f, 0.9f, 1.0f), Vec3f(1.1f, 1.0f, 0.9f)),
                        dnnBackendsAndTargetsWithHalide()
));

typedef TestWithParam<tuple<Vec3f, tuple<Backend, Target> > > Exp;
TEST_P(Exp, Accuracy)
{
    float base = get<0>(GetParam())[0];
    float scale = get<0>(GetParam())[1];
    float shift = get<0>(GetParam())[2];
    Backend backendId = get<0>(get<1>(GetParam()));
    Target targetId = get<1>(get<1>(GetParam()));

    LayerParams lp;
    lp.set("base", base);
    lp.set("scale", scale);
    lp.set("shift", shift);
    lp.type = "Exp";
    lp.name = "testLayer";
    testInPlaceActivation(lp, backendId, targetId);
}

INSTANTIATE_TEST_CASE_P(Layer_Test_Halide, Exp, Combine(
/*base, scale, shift*/ Values(Vec3f(0.9f, -1.0f, 1.1f), Vec3f(0.9f, 1.1f, -1.0f),
                              Vec3f(-1.0f, 0.9f, 1.1f), Vec3f(-1.0f, 1.1f, 0.9f),
                              Vec3f(1.1f, 0.9f, -1.0f), Vec3f(1.1f, -1.0f, 0.9f)),
                       dnnBackendsAndTargetsWithHalide()
));

TEST_P(Test_Halide_layers, ChannelsPReLU)
{
    LayerParams lp;
    lp.type = "ChannelsPReLU";
    lp.name = "testLayer";
    lp.blobs.push_back(Mat(1, kNumChannels, CV_32F));
    randu(lp.blobs[0], -1.0f, 1.0f);

    testInPlaceActivation(lp, backend, target);
}

typedef TestWithParam<tuple<bool, tuple<Backend, Target> > > Scale;
TEST_P(Scale, Accuracy)
{
    bool hasBias = get<0>(GetParam());
    Backend backendId = get<0>(get<1>(GetParam()));
    Target targetId = get<1>(get<1>(GetParam()));

    LayerParams lp;
    lp.set("bias_term", hasBias);
    lp.type = "Scale";
    lp.name = "testLayer";
    lp.blobs.push_back(Mat(1, kNumChannels, CV_32F));
    randu(lp.blobs[0], -1.0f, 1.0f);
    if (hasBias)
    {
        lp.blobs.push_back(Mat(1, kNumChannels, CV_32F));
        randu(lp.blobs[1], -1.0f, 1.0f);
    }
    testInPlaceActivation(lp, backendId, targetId);
}

INSTANTIATE_TEST_CASE_P(Layer_Test_Halide, Scale, Combine(
    Bool(),
    dnnBackendsAndTargetsWithHalide()
));

////////////////////////////////////////////////////////////////////////////////
// Concat layer
////////////////////////////////////////////////////////////////////////////////
//
// input --- conv --- concat --- output
//      `--- conv ----^ ^ ^
//      `---- ... ------' '
//      `-----------------'
typedef TestWithParam<tuple<Vec3i, Vec3i, tuple<Backend, Target> > > Concat;
TEST_P(Concat, Accuracy)
{
    Vec3i inSize = get<0>(GetParam());
    Vec3i numChannels = get<1>(GetParam());
    Backend backendId = get<0>(get<2>(GetParam()));
    Target targetId = get<1>(get<2>(GetParam()));

#if defined(INF_ENGINE_RELEASE) && INF_ENGINE_VER_MAJOR_LE(2018050000)
    if (backendId == DNN_BACKEND_INFERENCE_ENGINE_NN_BUILDER_2019 && targetId == DNN_TARGET_MYRIAD
            && inSize == Vec3i(1, 4, 5) && numChannels == Vec3i(1, 6, 2)
    )
        applyTestTag(CV_TEST_TAG_DNN_SKIP_IE_MYRIAD, CV_TEST_TAG_DNN_SKIP_IE_NN_BUILDER, CV_TEST_TAG_DNN_SKIP_IE_VERSION);  // crash
#endif

#if defined(INF_ENGINE_RELEASE) && INF_ENGINE_VER_MAJOR_GE(2019010000)
    if (backendId == DNN_BACKEND_INFERENCE_ENGINE_NN_BUILDER_2019 && targetId == DNN_TARGET_CPU
            && inSize == Vec3i(1, 4, 5) && numChannels == Vec3i(1, 6, 2)
    )
        applyTestTag(CV_TEST_TAG_DNN_SKIP_IE_NN_BUILDER, CV_TEST_TAG_DNN_SKIP_IE_VERSION);  // TODO: IE_CPU
#endif

    Net net;

    std::vector<int> convLayerIds;
    convLayerIds.reserve(numChannels.channels);
    for (int i = 0, n = numChannels.channels; i < n; ++i)
    {
        if (!numChannels[i])
            break;

        int sz[] = {numChannels[i], inSize[0], 1, 1};
        Mat weights(4, &sz[0], CV_32F);
        randu(weights, -1.0f, 1.0f);

        LayerParams convParam;
        convParam.set("kernel_w", 1);
        convParam.set("kernel_h", 1);
        convParam.set("num_output", numChannels[i]);
        convParam.set("bias_term", false);
        convParam.type = "Convolution";
        std::ostringstream ss;
        ss << "convLayer" << i;
        convParam.name = ss.str();
        convParam.blobs.push_back(weights);

        int layerId = net.addLayer(convParam.name, convParam.type, convParam);
        convLayerIds.push_back(layerId);
        net.connect(0, 0, layerId, 0);
    }

    LayerParams concatParam;
    concatParam.type = "Concat";
    concatParam.name = "testLayer";
    int concatId = net.addLayer(concatParam.name, concatParam.type, concatParam);
    net.connect(0, 0, concatId, 0);
    for (int i = 0; i < convLayerIds.size(); ++i)
    {
        net.connect(convLayerIds[i], 0, concatId, i + 1);
    }

    int sz[] = {1, inSize[0], inSize[1], inSize[2]};
    Mat input(4, &sz[0], CV_32F);
    test(input, net, backendId, targetId);
}

INSTANTIATE_TEST_CASE_P(Layer_Test_Halide, Concat, Combine(
/*input size*/ Values(Vec3i(1, 4, 5), Vec3i(2, 8, 6)),
/*channels*/   Values(Vec3i(2, 0, 0), Vec3i(3, 4, 0), Vec3i(1, 6, 2)),
               dnnBackendsAndTargetsWithHalide()
));

////////////////////////////////////////////////////////////////////////////////
// Element-wise layers
////////////////////////////////////////////////////////////////////////////////
//
// input --- conv --- eltwise --- output
//      `--- conv ----^ ^ ^
//      `---- ... ------' '
//      `-----------------'
typedef TestWithParam<tuple<Vec3i, std::string, int, bool, tuple<Backend, Target> > > Eltwise;
TEST_P(Eltwise, Accuracy)
{
    Vec3i inSize = get<0>(GetParam());
    std::string op = get<1>(GetParam());
    int numConv = get<2>(GetParam());
    bool weighted = get<3>(GetParam());
    Backend backendId = get<0>(get<4>(GetParam()));
    Target targetId = get<1>(get<4>(GetParam()));

#if defined(INF_ENGINE_RELEASE) && INF_ENGINE_VER_MAJOR_LE(2018050000)
    if (backendId == DNN_BACKEND_INFERENCE_ENGINE_NN_BUILDER_2019 && targetId == DNN_TARGET_MYRIAD &&
        inSize == Vec3i(1, 4, 5))
        applyTestTag(CV_TEST_TAG_DNN_SKIP_IE_MYRIAD, CV_TEST_TAG_DNN_SKIP_IE_NN_BUILDER, CV_TEST_TAG_DNN_SKIP_IE_VERSION);
#endif

#if defined(INF_ENGINE_RELEASE) && INF_ENGINE_VER_MAJOR_GE(2019010000)
    if (backendId == DNN_BACKEND_INFERENCE_ENGINE_NN_BUILDER_2019 && numConv > 1)
        applyTestTag(CV_TEST_TAG_DNN_SKIP_IE_NN_BUILDER, CV_TEST_TAG_DNN_SKIP_IE_VERSION);
#endif

#if defined(INF_ENGINE_RELEASE)
    if (backendId == DNN_BACKEND_INFERENCE_ENGINE_NN_BUILDER_2019 && targetId == DNN_TARGET_OPENCL &&
        op == "sum" && numConv == 1 && !weighted)
        applyTestTag(CV_TEST_TAG_DNN_SKIP_IE_OPENCL, CV_TEST_TAG_DNN_SKIP_IE_NN_BUILDER);
#endif

#if defined(INF_ENGINE_RELEASE)
    if (backendId == DNN_BACKEND_INFERENCE_ENGINE_NGRAPH && numConv > 1)
        applyTestTag(CV_TEST_TAG_DNN_SKIP_IE_NGRAPH, CV_TEST_TAG_DNN_SKIP_IE_VERSION);
#endif

    bool convInputShift = 1;
    int numEltwiseInputs = numConv;
    if (op == "div")
    {
        numConv = 1;
        convInputShift = 0; // first input is convolution
    }

    Net net;

    std::vector<int> convLayerIds(numConv);
    for (int i = 0; i < numConv; ++i)
    {
        int sz[] = {inSize[0], inSize[0], 1, 1};
        Mat weights(4, &sz[0], CV_32F);
        randu(weights, -1.0f, 1.0f);

        LayerParams convParam;
        convParam.set("kernel_w", 1);
        convParam.set("kernel_h", 1);
        convParam.set("num_output", inSize[0]);
        convParam.set("bias_term", false);
        convParam.type = "Convolution";
        std::ostringstream ss;
        ss << "convLayer" << i;
        convParam.name = ss.str();
        convParam.blobs.push_back(weights);

        convLayerIds[i] = net.addLayer(convParam.name, convParam.type, convParam);
        net.connect(0, 0, convLayerIds[i], 0);
    }

    LayerParams eltwiseParam;
    eltwiseParam.set("operation", op);
    if (op == "sum" && weighted)
    {
        RNG& rng = cv::theRNG();
        std::vector<float> coeff(1 + numConv);
        for (int i = 0; i < coeff.size(); ++i)
        {
            coeff[i] = rng.uniform(-2.0f, 2.0f);
        }
        eltwiseParam.set("coeff", DictValue::arrayReal<float*>(&coeff[0], coeff.size()));
    }
    eltwiseParam.type = "Eltwise";
    eltwiseParam.name = "testLayer";
    int eltwiseId = net.addLayer(eltwiseParam.name, eltwiseParam.type, eltwiseParam);
    if (convInputShift == 1)
        net.connect(0, 0, eltwiseId, 0);
    for (int i = 0; i < numConv; ++i)
    {
        net.connect(convLayerIds[i], 0, eltwiseId, i + convInputShift);
    }
    if (convInputShift == 0)
        net.connect(0, 0, eltwiseId, numConv);
    for (int i = numConv; i < numEltwiseInputs; ++i)
    {
        net.connect(0, 0, eltwiseId, i + 1);
    }

    int sz[] = {1, inSize[0], inSize[1], inSize[2]};
    Mat input(4, &sz[0], CV_32F);
    if (op == "div")
        randu(input, 1.0f, 1.0f);  // ensure no divisor value has absouluate value of less than 0.5
    test(input, net, backendId, targetId, /*skipCheck*/false, (op == "div") ? false : true);
}

INSTANTIATE_TEST_CASE_P(Layer_Test_Halide, Eltwise, Combine(
/*input size*/ Values(Vec3i(1, 4, 5), Vec3i(2, 8, 6)),
/*operation*/  Values("prod", "sum", "div", "max", "min"),
/*num convs*/  Values(1, 2, 3),
/*weighted(for sum only)*/ Bool(),
               dnnBackendsAndTargetsWithHalide()
));

////////////////////////////////////////////////////////////////////////////
// Mixed backends
////////////////////////////////////////////////////////////////////////////
#ifdef HAVE_HALIDE
TEST(MixedBackends_Halide_Default_Halide, Accuracy)
{
    // Just a layer that supports Halide backend.
    LayerParams lrn;
    lrn.type = "LRN";
    lrn.name = "testLRN";

    // Some of layers that doesn't supports Halide backend yet.
    LayerParams mvn;
    mvn.type = "MVN";
    mvn.name = "testMVN";

    // Halide layer again.
    LayerParams lrn2;
    lrn2.type = "LRN";
    lrn2.name = "testLRN2";

    Net net;
    int lrnId = net.addLayer(lrn.name, lrn.type, lrn);
    net.connect(0, 0, lrnId, 0);
    net.addLayerToPrev(mvn.name, mvn.type, mvn);
    net.addLayerToPrev(lrn2.name, lrn2.type, lrn2);

    int sz[] = {4, 3, 5, 6};
    Mat input(4, &sz[0], CV_32F);
    randu(input, -1.0f, 1.0f);
    net.setInput(input);
    net.setPreferableBackend(DNN_BACKEND_OPENCV);
    Mat outputDefault = net.forward().clone();

    net.setPreferableBackend(DNN_BACKEND_HALIDE);
    net.setInput(input);
    Mat outputHalide = net.forward().clone();
    normAssert(outputDefault, outputHalide);

    net.setPreferableTarget(DNN_TARGET_OPENCL);
    net.setInput(input);
    outputHalide = net.forward().clone();
    normAssert(outputDefault, outputHalide);
}
#endif  // HAVE_HALIDE

INSTANTIATE_TEST_CASE_P(/*nothing*/, Test_Halide_layers, dnnBackendsAndTargetsWithHalide());

}} // namespace