From c9564cb3850b6675cef663d7cc0722567b55cc25 Mon Sep 17 00:00:00 2001 From: Pablo Tello Date: Fri, 13 Sep 2019 10:20:25 +0100 Subject: COMPMID-2257: Implement NEGenerateProposals. Change-Id: I8d751f8b09f842a214c305a4530a71d82f16db8f Signed-off-by: Pablo Tello Reviewed-on: https://review.mlplatform.org/c/1943 Tested-by: Arm Jenkins Comments-Addressed: Arm Jenkins Reviewed-by: Michele Di Giorgio --- tests/validation/NEON/GenerateProposalsLayer.cpp | 403 +++++++++++++++++++++ .../validation/fixtures/ComputeAllAnchorsFixture.h | 2 +- 2 files changed, 404 insertions(+), 1 deletion(-) create mode 100644 tests/validation/NEON/GenerateProposalsLayer.cpp (limited to 'tests') diff --git a/tests/validation/NEON/GenerateProposalsLayer.cpp b/tests/validation/NEON/GenerateProposalsLayer.cpp new file mode 100644 index 0000000000..ea99bb3107 --- /dev/null +++ b/tests/validation/NEON/GenerateProposalsLayer.cpp @@ -0,0 +1,403 @@ +/* + * Copyright (c) 2019 ARM Limited. + * + * SPDX-License-Identifier: MIT + * + * Permission is hereby granted, free of charge, to any person obtaining a copy + * of this software and associated documentation files (the "Software"), to + * deal in the Software without restriction, including without limitation the + * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or + * sell copies of the Software, and to permit persons to whom the Software is + * furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in all + * copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE + * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER + * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, + * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE + * SOFTWARE. + */ +#include "arm_compute/runtime/NEON/NEScheduler.h" +#include "arm_compute/runtime/NEON/functions/NEComputeAllAnchors.h" +#include "arm_compute/runtime/NEON/functions/NEGenerateProposalsLayer.h" +#include "arm_compute/runtime/NEON/functions/NEPermute.h" +#include "arm_compute/runtime/NEON/functions/NESlice.h" +#include "tests/Globals.h" +#include "tests/NEON/Accessor.h" +#include "tests/NEON/ArrayAccessor.h" +#include "tests/framework/Macros.h" +#include "tests/framework/datasets/Datasets.h" +#include "tests/validation/Validation.h" +#include "tests/validation/fixtures/ComputeAllAnchorsFixture.h" +#include "utils/TypePrinter.h" + +namespace arm_compute +{ +namespace test +{ +namespace validation +{ +namespace +{ +template +inline void fill_tensor(U &&tensor, const std::vector &v) +{ + std::memcpy(tensor.data(), v.data(), sizeof(T) * v.size()); +} + +template +inline void fill_tensor(Accessor &&tensor, const std::vector &v) +{ + if(tensor.data_layout() == DataLayout::NCHW) + { + std::memcpy(tensor.data(), v.data(), sizeof(T) * v.size()); + } + else + { + const int channels = tensor.shape()[0]; + const int width = tensor.shape()[1]; + const int height = tensor.shape()[2]; + for(int x = 0; x < width; ++x) + { + for(int y = 0; y < height; ++y) + { + for(int c = 0; c < channels; ++c) + { + *(reinterpret_cast(tensor(Coordinates(c, x, y)))) = *(reinterpret_cast(v.data() + x + y * width + c * height * width)); + } + } + } + } +} + +const auto ComputeAllInfoDataset = framework::dataset::make("ComputeAllInfo", +{ + ComputeAnchorsInfo(10U, 10U, 1. / 16.f), + ComputeAnchorsInfo(100U, 1U, 1. / 2.f), + ComputeAnchorsInfo(100U, 1U, 1. / 4.f), + ComputeAnchorsInfo(100U, 100U, 1. / 4.f), + +}); +} // namespace + +TEST_SUITE(NEON) +TEST_SUITE(GenerateProposals) + +// *INDENT-OFF* +// clang-format off +DATA_TEST_CASE(Validate, framework::DatasetMode::ALL, zip(zip(zip(zip(zip(zip(zip( + framework::dataset::make("scores", { TensorInfo(TensorShape(100U, 100U, 9U), 1, DataType::F32), + TensorInfo(TensorShape(100U, 100U, 9U), 1, DataType::F16), // Mismatching types + TensorInfo(TensorShape(100U, 100U, 9U), 1, DataType::F16), // Wrong deltas (number of transformation non multiple of 4) + TensorInfo(TensorShape(100U, 100U, 9U), 1, DataType::F16), // Wrong anchors (number of values per roi != 5) + TensorInfo(TensorShape(100U, 100U, 9U), 1, DataType::F16), // Output tensor num_valid_proposals not scalar + TensorInfo(TensorShape(100U, 100U, 9U), 1, DataType::F16)}), // num_valid_proposals not U32 + framework::dataset::make("deltas",{ TensorInfo(TensorShape(100U, 100U, 36U), 1, DataType::F32), + TensorInfo(TensorShape(100U, 100U, 36U), 1, DataType::F32), + TensorInfo(TensorShape(100U, 100U, 38U), 1, DataType::F32), + TensorInfo(TensorShape(100U, 100U, 38U), 1, DataType::F32), + TensorInfo(TensorShape(100U, 100U, 38U), 1, DataType::F32), + TensorInfo(TensorShape(100U, 100U, 38U), 1, DataType::F32)})), + framework::dataset::make("anchors", { TensorInfo(TensorShape(4U, 9U), 1, DataType::F32), + TensorInfo(TensorShape(4U, 9U), 1, DataType::F32), + TensorInfo(TensorShape(4U, 9U), 1, DataType::F32), + TensorInfo(TensorShape(5U, 9U), 1, DataType::F32), + TensorInfo(TensorShape(4U, 9U), 1, DataType::F32), + TensorInfo(TensorShape(4U, 9U), 1, DataType::F32)})), + framework::dataset::make("proposals", { TensorInfo(TensorShape(5U, 100U*100U*9U), 1, DataType::F32), + TensorInfo(TensorShape(5U, 100U*100U*9U), 1, DataType::F32), + TensorInfo(TensorShape(5U, 100U*100U*9U), 1, DataType::F32), + TensorInfo(TensorShape(5U, 100U*100U*9U), 1, DataType::F32), + TensorInfo(TensorShape(5U, 100U*100U*9U), 1, DataType::F32), + TensorInfo(TensorShape(5U, 100U*100U*9U), 1, DataType::F32)})), + framework::dataset::make("scores_out", { TensorInfo(TensorShape(100U*100U*9U), 1, DataType::F32), + TensorInfo(TensorShape(100U*100U*9U), 1, DataType::F32), + TensorInfo(TensorShape(100U*100U*9U), 1, DataType::F32), + TensorInfo(TensorShape(100U*100U*9U), 1, DataType::F32), + TensorInfo(TensorShape(100U*100U*9U), 1, DataType::F32), + TensorInfo(TensorShape(100U*100U*9U), 1, DataType::F32)})), + framework::dataset::make("num_valid_proposals", { TensorInfo(TensorShape(1U, 1U), 1, DataType::U32), + TensorInfo(TensorShape(1U, 1U), 1, DataType::U32), + TensorInfo(TensorShape(1U, 1U), 1, DataType::U32), + TensorInfo(TensorShape(1U, 1U), 1, DataType::U32), + TensorInfo(TensorShape(1U, 10U), 1, DataType::U32), + TensorInfo(TensorShape(1U, 1U), 1, DataType::F16)})), + framework::dataset::make("generate_proposals_info", { GenerateProposalsInfo(10.f, 10.f, 1.f), + GenerateProposalsInfo(10.f, 10.f, 1.f), + GenerateProposalsInfo(10.f, 10.f, 1.f), + GenerateProposalsInfo(10.f, 10.f, 1.f), + GenerateProposalsInfo(10.f, 10.f, 1.f), + GenerateProposalsInfo(10.f, 10.f, 1.f)})), + framework::dataset::make("Expected", { true, false, false, false, false, false })), + scores, deltas, anchors, proposals, scores_out, num_valid_proposals, generate_proposals_info, expected) +{ + ARM_COMPUTE_EXPECT(bool(NEGenerateProposalsLayer::validate(&scores.clone()->set_is_resizable(true), + &deltas.clone()->set_is_resizable(true), + &anchors.clone()->set_is_resizable(true), + &proposals.clone()->set_is_resizable(true), + &scores_out.clone()->set_is_resizable(true), + &num_valid_proposals.clone()->set_is_resizable(true), + generate_proposals_info)) == expected, framework::LogLevel::ERRORS); +} +// clang-format on +// *INDENT-ON* + +template +using NEComputeAllAnchorsFixture = ComputeAllAnchorsFixture; + +TEST_SUITE(Float) +TEST_SUITE(FP32) +DATA_TEST_CASE(IntegrationTestCaseAllAnchors, framework::DatasetMode::ALL, framework::dataset::make("DataType", { DataType::F32 }), + data_type) +{ + const int values_per_roi = 4; + const int num_anchors = 3; + const int feature_height = 4; + const int feature_width = 3; + + SimpleTensor anchors_expected(TensorShape(values_per_roi, feature_width * feature_height * num_anchors), DataType::F32); + fill_tensor(anchors_expected, std::vector { -26, -19, 87, 86, + -81, -27, 58, 63, + -44, -15, 55, 36, + -10, -19, 103, 86, + -65, -27, 74, 63, + -28, -15, 71, 36, + 6, -19, 119, 86, + -49, -27, 90, 63, + -12, -15, 87, 36, + -26, -3, 87, 102, + -81, -11, 58, 79, + -44, 1, 55, 52, + -10, -3, 103, 102, + -65, -11, 74, 79, + -28, 1, 71, 52, + 6, -3, 119, 102, + -49, -11, 90, 79, + -12, 1, 87, 52, + -26, 13, 87, 118, + -81, 5, 58, 95, + -44, 17, 55, 68, + -10, 13, 103, 118, + -65, 5, 74, 95, + -28, 17, 71, 68, + 6, 13, 119, 118, + -49, 5, 90, 95, + -12, 17, 87, 68, + -26, 29, 87, 134, + -81, 21, 58, 111, + -44, 33, 55, 84, + -10, 29, 103, 134, + -65, 21, 74, 111, + -28, 33, 71, 84, + 6, 29, 119, 134, + -49, 21, 90, 111, + -12, 33, 87, 84 + }); + + Tensor all_anchors; + Tensor anchors = create_tensor(TensorShape(4, num_anchors), data_type); + + // Create and configure function + NEComputeAllAnchors compute_anchors; + compute_anchors.configure(&anchors, &all_anchors, ComputeAnchorsInfo(feature_width, feature_height, 1. / 16.0)); + anchors.allocator()->allocate(); + all_anchors.allocator()->allocate(); + + fill_tensor(Accessor(anchors), std::vector { -26, -19, 87, 86, + -81, -27, 58, 63, + -44, -15, 55, 36 + }); + // Compute function + compute_anchors.run(); + validate(Accessor(all_anchors), anchors_expected); +} + +DATA_TEST_CASE(IntegrationTestCaseGenerateProposals, framework::DatasetMode::ALL, combine(framework::dataset::make("DataType", { DataType::F32 }), + framework::dataset::make("DataLayout", { DataLayout::NCHW, DataLayout::NHWC })), + data_type, data_layout) +{ + const int values_per_roi = 4; + const int num_anchors = 2; + const int feature_height = 4; + const int feature_width = 5; + + std::vector scores_vector + { + 5.055894435664012e-04f, 1.270304909820112e-03f, 2.492271113912067e-03f, 5.951663827809190e-03f, + 7.846917156877404e-03f, 6.776275276294789e-03f, 6.761571012891965e-03f, 4.898292096237725e-03f, + 6.044472332578605e-04f, 3.203334118759474e-03f, 2.947527908919908e-03f, 6.313238560015770e-03f, + 7.931767757095738e-03f, 8.764345805102866e-03f, 7.325012199914913e-03f, 4.317069470446271e-03f, + 2.372537409795522e-03f, 1.589227460352735e-03f, 7.419477503600818e-03f, 3.157690354133824e-05f, + 1.125915135986472e-03f, 9.865363483872330e-03f, 2.429454743386769e-03f, 2.724460564167563e-03f, + 7.670409838207963e-03f, 5.558891552328172e-03f, 7.876904873099614e-03f, 6.824746047239291e-03f, + 7.023817548067892e-03f, 3.651314909238673e-04f, 6.720443709032501e-03f, 5.935615511606155e-03f, + 2.837349642759774e-03f, 1.787235113610299e-03f, 4.538568889918262e-03f, 3.391510678188818e-03f, + 7.328474239481874e-03f, 6.306967923936016e-03f, 8.102218904895860e-04f, 3.366646521610209e-03f + }; + + std::vector bbx_vector + { + 5.066650471856862e-03, -7.638671742936328e-03, 2.549596503988635e-03, -8.316416756423296e-03, + -2.397471917924575e-04, 7.370595187754891e-03, -2.771880178185262e-03, 3.958364873973579e-03, + 4.493661094712284e-03, 2.016487051533088e-03, -5.893883038142033e-03, 7.570636080807809e-03, + -1.395511229386785e-03, 3.686686052704696e-03, -7.738166245767079e-03, -1.947306329828059e-03, + -9.299719716045681e-03, -3.476410493413708e-03, -2.390761190919604e-03, 4.359281254364210e-03, + -2.135251160164030e-04, 9.203299843371962e-03, 4.042322775006053e-03, -9.464271243910754e-03, + 2.566239543229305e-03, -9.691093900220627e-03, -4.019283034310979e-03, 8.145470429508792e-03, + 7.345087308315662e-04, 7.049642787384043e-03, -2.768492313674294e-03, 6.997160053405803e-03, + 6.675346697112969e-03, 2.353293365652274e-03, -3.612002585241749e-04, 1.592076522068768e-03, + -8.354188900818149e-04, -5.232515333564140e-04, 6.946683728847089e-03, -8.469757407935994e-03, + -8.985324496496555e-03, 4.885832859017961e-03, -7.662967577576512e-03, 7.284124004335807e-03, + -5.812167510299458e-03, -5.760336800482398e-03, 6.040416930336549e-03, 5.861508595443691e-03, + -5.509243096133549e-04, -2.006142470055888e-03, -7.205925340416066e-03, -1.117459082969758e-03, + 4.233247017623154e-03, 8.079257498201178e-03, 2.962639022639513e-03, 7.069474943472751e-03, + -8.562946284971293e-03, -8.228634642768271e-03, -6.116245322799971e-04, -7.213122000180859e-03, + 1.693094399433209e-03, -4.287504459132290e-03, 8.740365683925144e-03, 3.751788160720638e-03, + 7.006764222862830e-03, 9.676754678358187e-03, -6.458757235812945e-03, -4.486506575589758e-03, + -4.371087196816259e-03, 3.542166755953152e-03, -2.504808998699504e-03, 5.666601724512010e-03, + -3.691862724546129e-03, 3.689809719085287e-03, 9.079930264704458e-03, 6.365127787359476e-03, + 2.881681788246101e-06, 9.991866069315165e-03, -1.104757466496565e-03, -2.668455405633477e-03, + -1.225748887087659e-03, 6.530536159094015e-03, 3.629468917975644e-03, 1.374426066950348e-03, + -2.404098881570632e-03, -4.791365049441602e-03, -2.970654027009094e-03, 7.807553690294366e-03, + -1.198321129505323e-03, -3.574885336949881e-03, -5.380848303732298e-03, 9.705151282165116e-03, + -1.005217683242201e-03, 9.178094036278405e-03, -5.615977269541644e-03, 5.333533158509859e-03, + -2.817116206168516e-03, 6.672609782000503e-03, 6.575769501651313e-03, 8.987596634989362e-03, + -1.283530791296188e-03, 1.687717120057778e-03, 3.242391851439037e-03, -7.312060454341677e-03, + 4.735335326324270e-03, -6.832367028817463e-03, -5.414854835884652e-03, -9.352380213755996e-03, + -3.682662043703889e-03, -6.127508590419776e-04, -7.682256596819467e-03, 9.569532628790246e-03, + -1.572157284518933e-03, -6.023034366859191e-03, -5.110873282582924e-03, -8.697072236660256e-03, + -3.235150419663566e-03, -8.286320236471386e-03, -5.229472409112913e-03, 9.920785896115053e-03, + -2.478413362126123e-03, -9.261324796935007e-03, 1.718512310840434e-04, 3.015875488208480e-03, + -6.172932549255669e-03, -4.031715551985103e-03, -9.263878005853677e-03, -2.815310738453385e-03, + 7.075307462133643e-03, 1.404611747938669e-03, -1.518548732533266e-03, -9.293430941655778e-03, + 6.382186966633246e-03, 8.256835789169248e-03, 3.196907843506736e-03, 8.821615689753433e-03, + -7.661543424832439e-03, 1.636273081822326e-03, -8.792373335756125e-03, 2.958775812049877e-03, + -6.269300278071262e-03, 6.248285790856450e-03, -3.675414624536002e-03, -1.692616700318762e-03, + 4.126007647815893e-03, -9.155291689759584e-03, -8.432616039924004e-03, 4.899980636213323e-03, + 3.511535019681671e-03, -1.582745757177339e-03, -2.703657774917963e-03, 6.738168990840388e-03, + 4.300455303937919e-03, 9.618312854781494e-03, 2.762142918402472e-03, -6.590025003382154e-03, + -2.071168373801788e-03, 8.613893943683627e-03, 9.411190295341036e-03, -6.129018930548372e-03 + }; + + const std::vector anchors_vector{ -26, -19, 87, 86, -81, -27, 58, 63 }; + ; + + SimpleTensor proposals_expected(TensorShape(5, 9), DataType::F32); + fill_tensor(proposals_expected, std::vector + { + 0, 0, 0, 75.269, 64.4388, + 0, 21.9579, 13.0535, 119, 99, + 0, 38.303, 0, 119, 87.6447, + 0, 0, 0, 119, 64.619, + 0, 0, 20.7997, 74.0714, 99, + 0, 0, 0, 91.8963, 79.3724, + 0, 0, 4.42377, 58.1405, 95.1781, + 0, 0, 13.4405, 104.799, 99, + 0, 38.9066, 28.2434, 119, 99, + + }); + + SimpleTensor scores_expected(TensorShape(9), DataType::F32); + fill_tensor(scores_expected, std::vector + { + 0.00986536, + 0.00876435, + 0.00784692, + 0.00767041, + 0.00732847, + 0.00682475, + 0.00672044, + 0.00631324, + 3.15769e-05 + }); + + TensorShape scores_shape = TensorShape(feature_width, feature_height, num_anchors); + TensorShape deltas_shape = TensorShape(feature_width, feature_height, values_per_roi * num_anchors); + if(data_layout == DataLayout::NHWC) + { + permute(scores_shape, PermutationVector(2U, 0U, 1U)); + permute(deltas_shape, PermutationVector(2U, 0U, 1U)); + } + // Inputs + Tensor scores = create_tensor(scores_shape, data_type, 1, QuantizationInfo(), data_layout); + Tensor bbox_deltas = create_tensor(deltas_shape, data_type, 1, QuantizationInfo(), data_layout); + Tensor anchors = create_tensor(TensorShape(values_per_roi, num_anchors), data_type); + + // Outputs + Tensor proposals; + Tensor num_valid_proposals; + Tensor scores_out; + num_valid_proposals.allocator()->init(TensorInfo(TensorShape(1), 1, DataType::U32)); + + NEGenerateProposalsLayer generate_proposals; + generate_proposals.configure(&scores, &bbox_deltas, &anchors, &proposals, &scores_out, &num_valid_proposals, + GenerateProposalsInfo(120, 100, 0.166667f, 1 / 16.0, 6000, 300, 0.7f, 16.0f)); + + // Allocate memory for input/output tensors + scores.allocator()->allocate(); + bbox_deltas.allocator()->allocate(); + anchors.allocator()->allocate(); + proposals.allocator()->allocate(); + num_valid_proposals.allocator()->allocate(); + scores_out.allocator()->allocate(); + // Fill inputs + fill_tensor(Accessor(scores), scores_vector); + fill_tensor(Accessor(bbox_deltas), bbx_vector); + fill_tensor(Accessor(anchors), anchors_vector); + + // Run operator + generate_proposals.run(); + // Gather num_valid_proposals + const uint32_t N = *reinterpret_cast(num_valid_proposals.ptr_to_element(Coordinates(0, 0))); + + // Select the first N entries of the proposals + Tensor proposals_final; + NESlice select_proposals; + select_proposals.configure(&proposals, &proposals_final, Coordinates(0, 0), Coordinates(values_per_roi + 1, N)); + + proposals_final.allocator()->allocate(); + select_proposals.run(); + + // Select the first N entries of the proposals + Tensor scores_final; + NESlice select_scores; + select_scores.configure(&scores_out, &scores_final, Coordinates(0), Coordinates(N)); + scores_final.allocator()->allocate(); + select_scores.run(); + + const RelativeTolerance tolerance_f32(1e-5f); + // Validate the output + validate(Accessor(proposals_final), proposals_expected, tolerance_f32); + validate(Accessor(scores_final), scores_expected, tolerance_f32); +} + +FIXTURE_DATA_TEST_CASE(ComputeAllAnchors, NEComputeAllAnchorsFixture, framework::DatasetMode::ALL, + combine(combine(framework::dataset::make("NumAnchors", { 2, 4, 8 }), ComputeAllInfoDataset), framework::dataset::make("DataType", { DataType::F32 }))) +{ + // Validate output + validate(Accessor(_target), _reference); +} +TEST_SUITE_END() // FP32 +#ifdef __ARM_FEATURE_FP16_VECTOR_ARITHMETIC +TEST_SUITE(FP16) +FIXTURE_DATA_TEST_CASE(ComputeAllAnchors, NEComputeAllAnchorsFixture, framework::DatasetMode::ALL, + combine(combine(framework::dataset::make("NumAnchors", { 2, 4, 8 }), ComputeAllInfoDataset), framework::dataset::make("DataType", { DataType::F16 }))) +{ + // Validate output + validate(Accessor(_target), _reference); +} +TEST_SUITE_END() // FP16 +#endif // __ARM_FEATURE_FP16_VECTOR_ARITHMETIC + +TEST_SUITE_END() // Float + +TEST_SUITE_END() // GenerateProposals +TEST_SUITE_END() // NEON + +} // namespace validation +} // namespace test +} // namespace arm_compute diff --git a/tests/validation/fixtures/ComputeAllAnchorsFixture.h b/tests/validation/fixtures/ComputeAllAnchorsFixture.h index bfa43ceafc..6f2db3e623 100644 --- a/tests/validation/fixtures/ComputeAllAnchorsFixture.h +++ b/tests/validation/fixtures/ComputeAllAnchorsFixture.h @@ -78,7 +78,7 @@ protected: ARM_COMPUTE_EXPECT(!all_anchors.info()->is_resizable(), framework::LogLevel::ERRORS); // Fill tensors - fill(CLAccessor(anchors)); + fill(AccessorType(anchors)); // Compute function compute_all_anchors.run(); -- cgit v1.2.1