COMPMID-496: Port DepthConvert to new validation

Change-Id: I4b5faeadcaea2878687e0c3844f5fe47fd0a3c8d
Reviewed-on: http://mpd-gerrit.cambridge.arm.com/87320
Tested-by: Kaizen <jeremy.johnson+kaizengerrit@arm.com>
Reviewed-by: Moritz Pflanzer <moritz.pflanzer@arm.com>

5 files changed, 1315 insertions, 0 deletions

diff --git a/tests/validation/CL/DepthConvert.cpp b/tests/validation/CL/DepthConvert.cpp
new file mode 100644
index 0000000000..6abb482f49
--- /dev/null
+++ b/tests/validation/CL/DepthConvert.cpp
@@ -0,0 +1,485 @@
+/*
+ * Copyright (c) 2017 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 CONCLCTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+#include "arm_compute/core/Types.h"
+#include "arm_compute/runtime/CL/CLTensor.h"
+#include "arm_compute/runtime/CL/CLTensorAllocator.h"
+#include "arm_compute/runtime/CL/functions/CLDepthConvert.h"
+#include "tests/CL/CLAccessor.h"
+#include "tests/PaddingCalculator.h"
+#include "tests/datasets/ConvertPolicyDataset.h"
+#include "tests/datasets/ShapeDatasets.h"
+#include "tests/framework/Asserts.h"
+#include "tests/framework/Macros.h"
+#include "tests/framework/datasets/Datasets.h"
+#include "tests/validation/Validation.h"
+#include "tests/validation/fixtures/DepthConvertFixture.h"
+#include "tests/validation/half.h"
+
+namespace arm_compute
+{
+namespace test
+{
+namespace validation
+{
+namespace
+{
+/** Input data sets **/
+const auto DepthConvertU8toU16Dataset             = combine(framework::dataset::make("DataType", DataType::U8), framework::dataset::make("DataType", DataType::U16));
+const auto DepthConvertU8toS16Dataset             = combine(framework::dataset::make("DataType", DataType::U8), framework::dataset::make("DataType", DataType::S16));
+const auto DepthConvertU8toS32Dataset             = combine(framework::dataset::make("DataType", DataType::U8), framework::dataset::make("DataType", DataType::S32));
+const auto DepthConvertU16toU8Dataset             = combine(framework::dataset::make("DataType", DataType::U16), framework::dataset::make("DataType", DataType::U8));
+const auto DepthConvertU16toU32Dataset            = combine(framework::dataset::make("DataType", DataType::U16), framework::dataset::make("DataType", DataType::U32));
+const auto DepthConvertS16toU8Dataset             = combine(framework::dataset::make("DataType", DataType::S16), framework::dataset::make("DataType", DataType::U8));
+const auto DepthConvertS16toS32Dataset            = combine(framework::dataset::make("DataType", DataType::S16), framework::dataset::make("DataType", DataType::S32));
+const auto DepthConvertQS8toFP32Dataset           = combine(framework::dataset::make("DataType", DataType::QS8), framework::dataset::make("DataType", DataType::F32));
+const auto DepthConvertQS16toFP32Dataset          = combine(framework::dataset::make("DataType", DataType::QS16), framework::dataset::make("DataType", DataType::F32));
+const auto DepthConvertFP32toQS8Dataset           = combine(framework::dataset::make("DataType", DataType::F32), framework::dataset::make("DataType", DataType::QS8));
+const auto DepthConvertFP32toQS16Dataset          = combine(framework::dataset::make("DataType", DataType::F32), framework::dataset::make("DataType", DataType::QS16));
+const auto DepthConvertShiftDataset               = framework::dataset::make("Shift", 0, 7);
+const auto DepthConvertFixedPointQuantizedDataset = framework::dataset::make("FractionalBits", 1, 7);
+} // namespace
+
+TEST_SUITE(CL)
+TEST_SUITE(DepthConvert)
+template <typename T>
+using CLDepthConvertToU16Fixture = DepthConvertValidationFixture<CLTensor, CLAccessor, CLDepthConvert, T, uint16_t>;
+template <typename T>
+using CLDepthConvertToS16Fixture = DepthConvertValidationFixture<CLTensor, CLAccessor, CLDepthConvert, T, int16_t>;
+template <typename T>
+using CLDepthConvertToS32Fixture = DepthConvertValidationFixture<CLTensor, CLAccessor, CLDepthConvert, T, int32_t>;
+template <typename T>
+using CLDepthConvertToU8Fixture = DepthConvertValidationFixture<CLTensor, CLAccessor, CLDepthConvert, T, uint8_t>;
+template <typename T>
+using CLDepthConvertToU32Fixture = DepthConvertValidationFixture<CLTensor, CLAccessor, CLDepthConvert, T, uint32_t>;
+template <typename T>
+using CLDepthConvertToFP32FixedPointFixture = DepthConvertValidationFractionalBitsFixture<CLTensor, CLAccessor, CLDepthConvert, T, float>;
+template <typename T>
+using CLDepthConvertToQS8FixedPointFixture = DepthConvertValidationFractionalBitsFixture<CLTensor, CLAccessor, CLDepthConvert, T, int8_t>;
+template <typename T>
+using CLDepthConvertToQS16FixedPointFixture = DepthConvertValidationFractionalBitsFixture<CLTensor, CLAccessor, CLDepthConvert, T, int16_t>;
+
+TEST_SUITE(U8_to_U16)
+DATA_TEST_CASE(Configuration, framework::DatasetMode::ALL, combine(combine(framework::dataset::concat(datasets::SmallShapes(), datasets::LargeShapes()), framework::dataset::make("ConvertPolicy", { ConvertPolicy::SATURATE, ConvertPolicy::WRAP })),
+                                                                   DepthConvertShiftDataset),
+               shape, policy, shift)
+{
+    int fixed_point_position = 0;
+
+    // Create tensors
+    CLTensor src = create_tensor<CLTensor>(shape, DataType::U8, 1, fixed_point_position);
+    CLTensor dst = create_tensor<CLTensor>(shape, DataType::U16, 1, fixed_point_position);
+
+    // Create and Configure function
+    CLDepthConvert depth_convert;
+    depth_convert.configure(&src, &dst, policy, shift);
+
+    // Validate valid region
+    const ValidRegion valid_region = shape_to_valid_region(shape);
+    validate(dst.info()->valid_region(), valid_region);
+
+    // Validate padding
+    const PaddingSize padding = PaddingCalculator(shape.x(), 16).required_padding();
+    validate(src.info()->padding(), padding);
+    validate(dst.info()->padding(), padding);
+}
+
+FIXTURE_DATA_TEST_CASE(RunSmall, CLDepthConvertToU16Fixture<uint8_t>, framework::DatasetMode::PRECOMMIT, combine(combine(combine(datasets::SmallShapes(), DepthConvertU8toU16Dataset),
+                                                                                                                 framework::dataset::make("ConvertPolicy", { ConvertPolicy::SATURATE, ConvertPolicy::WRAP })),
+                                                                                                                 DepthConvertShiftDataset))
+{
+    // Validate output
+    validate(CLAccessor(_target), _reference);
+}
+
+FIXTURE_DATA_TEST_CASE(RunLarge, CLDepthConvertToU16Fixture<uint8_t>, framework::DatasetMode::NIGHTLY, combine(combine(combine(datasets::LargeShapes(), DepthConvertU8toU16Dataset),
+                                                                                                                       framework::dataset::make("ConvertPolicy", { ConvertPolicy::SATURATE, ConvertPolicy::WRAP })),
+                                                                                                               DepthConvertShiftDataset))
+{
+    // Validate output
+    validate(CLAccessor(_target), _reference);
+}
+TEST_SUITE_END()
+
+TEST_SUITE(U8_to_S16)
+DATA_TEST_CASE(Configuration, framework::DatasetMode::ALL, combine(combine(framework::dataset::concat(datasets::SmallShapes(), datasets::LargeShapes()), framework::dataset::make("ConvertPolicy", { ConvertPolicy::SATURATE, ConvertPolicy::WRAP })),
+                                                                   DepthConvertShiftDataset),
+               shape, policy, shift)
+{
+    int fixed_point_position = 0;
+
+    // Create tensors
+    CLTensor src = create_tensor<CLTensor>(shape, DataType::U8, 1, fixed_point_position);
+    CLTensor dst = create_tensor<CLTensor>(shape, DataType::S16, 1, fixed_point_position);
+
+    // Create and Configure function
+    CLDepthConvert depth_convert;
+    depth_convert.configure(&src, &dst, policy, shift);
+
+    // Validate valid region
+    const ValidRegion valid_region = shape_to_valid_region(shape);
+    validate(dst.info()->valid_region(), valid_region);
+
+    // Validate padding
+    const PaddingSize padding = PaddingCalculator(shape.x(), 16).required_padding();
+    validate(src.info()->padding(), padding);
+    validate(dst.info()->padding(), padding);
+}
+
+FIXTURE_DATA_TEST_CASE(RunSmall, CLDepthConvertToS16Fixture<uint8_t>, framework::DatasetMode::PRECOMMIT, combine(combine(combine(datasets::SmallShapes(), DepthConvertU8toS16Dataset),
+                                                                                                                 framework::dataset::make("ConvertPolicy", { ConvertPolicy::SATURATE, ConvertPolicy::WRAP })),
+                                                                                                                 DepthConvertShiftDataset))
+{
+    // Validate output
+    validate(CLAccessor(_target), _reference);
+}
+
+FIXTURE_DATA_TEST_CASE(RunLarge, CLDepthConvertToS16Fixture<uint8_t>, framework::DatasetMode::NIGHTLY, combine(combine(combine(datasets::LargeShapes(), DepthConvertU8toS16Dataset),
+                                                                                                                       framework::dataset::make("ConvertPolicy", { ConvertPolicy::SATURATE, ConvertPolicy::WRAP })),
+                                                                                                               DepthConvertShiftDataset))
+{
+    // Validate output
+    validate(CLAccessor(_target), _reference);
+}
+TEST_SUITE_END()
+TEST_SUITE(U8_to_S32)
+DATA_TEST_CASE(Configuration, framework::DatasetMode::ALL, combine(combine(framework::dataset::concat(datasets::SmallShapes(), datasets::LargeShapes()), framework::dataset::make("ConvertPolicy", { ConvertPolicy::SATURATE, ConvertPolicy::WRAP })),
+                                                                   DepthConvertShiftDataset),
+               shape, policy, shift)
+{
+    int fixed_point_position = 0;
+
+    // Create tensors
+    CLTensor src = create_tensor<CLTensor>(shape, DataType::U8, 1, fixed_point_position);
+    CLTensor dst = create_tensor<CLTensor>(shape, DataType::S32, 1, fixed_point_position);
+
+    // Create and Configure function
+    CLDepthConvert depth_convert;
+    depth_convert.configure(&src, &dst, policy, shift);
+
+    // Validate valid region
+    const ValidRegion valid_region = shape_to_valid_region(shape);
+    validate(dst.info()->valid_region(), valid_region);
+
+    // Validate padding
+    const PaddingSize padding = PaddingCalculator(shape.x(), 16).required_padding();
+    validate(src.info()->padding(), padding);
+    validate(dst.info()->padding(), padding);
+}
+
+FIXTURE_DATA_TEST_CASE(RunSmall, CLDepthConvertToS32Fixture<uint8_t>, framework::DatasetMode::PRECOMMIT, combine(combine(combine(datasets::SmallShapes(), DepthConvertU8toS32Dataset),
+                                                                                                                 framework::dataset::make("ConvertPolicy", { ConvertPolicy::SATURATE, ConvertPolicy::WRAP })),
+                                                                                                                 DepthConvertShiftDataset))
+{
+    // Validate output
+    validate(CLAccessor(_target), _reference);
+}
+
+FIXTURE_DATA_TEST_CASE(RunLarge, CLDepthConvertToS32Fixture<uint8_t>, framework::DatasetMode::NIGHTLY, combine(combine(combine(datasets::LargeShapes(), DepthConvertU8toS32Dataset),
+                                                                                                                       framework::dataset::make("ConvertPolicy", { ConvertPolicy::SATURATE, ConvertPolicy::WRAP })),
+                                                                                                               DepthConvertShiftDataset))
+{
+    // Validate output
+    validate(CLAccessor(_target), _reference);
+}
+TEST_SUITE_END()
+
+TEST_SUITE(U16_to_U8)
+DATA_TEST_CASE(Configuration, framework::DatasetMode::ALL, combine(combine(framework::dataset::concat(datasets::SmallShapes(), datasets::LargeShapes()), framework::dataset::make("ConvertPolicy", { ConvertPolicy::SATURATE, ConvertPolicy::WRAP })),
+                                                                   DepthConvertShiftDataset),
+               shape, policy, shift)
+{
+    int fixed_point_position = 0;
+
+    // Create tensors
+    CLTensor src = create_tensor<CLTensor>(shape, DataType::U16, 1, fixed_point_position);
+    CLTensor dst = create_tensor<CLTensor>(shape, DataType::U8, 1, fixed_point_position);
+
+    // Create and Configure function
+    CLDepthConvert depth_convert;
+    depth_convert.configure(&src, &dst, policy, shift);
+
+    // Validate valid region
+    const ValidRegion valid_region = shape_to_valid_region(shape);
+    validate(dst.info()->valid_region(), valid_region);
+
+    // Validate padding
+    const PaddingSize padding = PaddingCalculator(shape.x(), 16).required_padding();
+    validate(src.info()->padding(), padding);
+    validate(dst.info()->padding(), padding);
+}
+
+FIXTURE_DATA_TEST_CASE(RunSmall, CLDepthConvertToU8Fixture<uint16_t>, framework::DatasetMode::PRECOMMIT, combine(combine(combine(datasets::SmallShapes(), DepthConvertU16toU8Dataset),
+                                                                                                                 framework::dataset::make("ConvertPolicy", { ConvertPolicy::SATURATE, ConvertPolicy::WRAP })),
+                                                                                                                 DepthConvertShiftDataset))
+{
+    // Validate output
+    validate(CLAccessor(_target), _reference);
+}
+FIXTURE_DATA_TEST_CASE(RunLarge, CLDepthConvertToU8Fixture<uint16_t>, framework::DatasetMode::NIGHTLY, combine(combine(combine(datasets::LargeShapes(), DepthConvertU16toU8Dataset),
+                                                                                                                       framework::dataset::make("ConvertPolicy", { ConvertPolicy::SATURATE, ConvertPolicy::WRAP })),
+                                                                                                               DepthConvertShiftDataset))
+{
+    // Validate output
+    validate(CLAccessor(_target), _reference);
+}
+TEST_SUITE_END()
+
+TEST_SUITE(U16_to_U32)
+DATA_TEST_CASE(Configuration, framework::DatasetMode::ALL, combine(combine(framework::dataset::concat(datasets::SmallShapes(), datasets::LargeShapes()), framework::dataset::make("ConvertPolicy", { ConvertPolicy::SATURATE, ConvertPolicy::WRAP })),
+                                                                   DepthConvertShiftDataset),
+               shape, policy, shift)
+{
+    int fixed_point_position = 0;
+
+    // Create tensors
+    CLTensor src = create_tensor<CLTensor>(shape, DataType::U16, 1, fixed_point_position);
+    CLTensor dst = create_tensor<CLTensor>(shape, DataType::U32, 1, fixed_point_position);
+
+    // Create and Configure function
+    CLDepthConvert depth_convert;
+    depth_convert.configure(&src, &dst, policy, shift);
+
+    // Validate valid region
+    const ValidRegion valid_region = shape_to_valid_region(shape);
+    validate(dst.info()->valid_region(), valid_region);
+
+    // Validate padding
+    const PaddingSize padding = PaddingCalculator(shape.x(), 16).required_padding();
+    validate(src.info()->padding(), padding);
+    validate(dst.info()->padding(), padding);
+}
+
+FIXTURE_DATA_TEST_CASE(RunSmall, CLDepthConvertToU32Fixture<uint16_t>, framework::DatasetMode::PRECOMMIT, combine(combine(combine(datasets::SmallShapes(), DepthConvertU16toU32Dataset),
+                                                                                                                  framework::dataset::make("ConvertPolicy", { ConvertPolicy::SATURATE, ConvertPolicy::WRAP })),
+                                                                                                                  DepthConvertShiftDataset))
+{
+    // Validate output
+    validate(CLAccessor(_target), _reference);
+}
+FIXTURE_DATA_TEST_CASE(RunLarge, CLDepthConvertToU32Fixture<uint16_t>, framework::DatasetMode::NIGHTLY, combine(combine(combine(datasets::LargeShapes(), DepthConvertU16toU32Dataset),
+                                                                                                                        framework::dataset::make("ConvertPolicy", { ConvertPolicy::SATURATE, ConvertPolicy::WRAP })),
+                                                                                                                DepthConvertShiftDataset))
+{
+    // Validate output
+    validate(CLAccessor(_target), _reference);
+}
+TEST_SUITE_END()
+
+TEST_SUITE(S16_to_U8)
+DATA_TEST_CASE(Configuration, framework::DatasetMode::ALL, combine(combine(framework::dataset::concat(datasets::SmallShapes(), datasets::LargeShapes()), framework::dataset::make("ConvertPolicy", { ConvertPolicy::SATURATE, ConvertPolicy::WRAP })),
+                                                                   DepthConvertShiftDataset),
+               shape, policy, shift)
+{
+    int fixed_point_position = 0;
+
+    // Create tensors
+    CLTensor src = create_tensor<CLTensor>(shape, DataType::S16, 1, fixed_point_position);
+    CLTensor dst = create_tensor<CLTensor>(shape, DataType::U8, 1, fixed_point_position);
+
+    // Create and Configure function
+    CLDepthConvert depth_convert;
+    depth_convert.configure(&src, &dst, policy, shift);
+
+    // Validate valid region
+    const ValidRegion valid_region = shape_to_valid_region(shape);
+    validate(dst.info()->valid_region(), valid_region);
+
+    // Validate padding
+    const PaddingSize padding = PaddingCalculator(shape.x(), 16).required_padding();
+    validate(src.info()->padding(), padding);
+    validate(dst.info()->padding(), padding);
+}
+
+FIXTURE_DATA_TEST_CASE(RunSmall, CLDepthConvertToU8Fixture<int16_t>, framework::DatasetMode::PRECOMMIT, combine(combine(combine(datasets::SmallShapes(), DepthConvertS16toU8Dataset),
+                                                                                                                        framework::dataset::make("ConvertPolicy", { ConvertPolicy::SATURATE, ConvertPolicy::WRAP })),
+                                                                                                                DepthConvertShiftDataset))
+{
+    // Validate output
+    validate(CLAccessor(_target), _reference);
+}
+FIXTURE_DATA_TEST_CASE(RunLarge, CLDepthConvertToU8Fixture<int16_t>, framework::DatasetMode::NIGHTLY, combine(combine(combine(datasets::LargeShapes(), DepthConvertS16toU8Dataset),
+                                                                                                                      framework::dataset::make("ConvertPolicy", { ConvertPolicy::SATURATE, ConvertPolicy::WRAP })),
+                                                                                                              DepthConvertShiftDataset))
+{
+    // Validate output
+    validate(CLAccessor(_target), _reference);
+}
+TEST_SUITE_END()
+
+TEST_SUITE(S16_to_S32)
+DATA_TEST_CASE(Configuration, framework::DatasetMode::ALL, combine(combine(framework::dataset::concat(datasets::SmallShapes(), datasets::LargeShapes()), framework::dataset::make("ConvertPolicy", { ConvertPolicy::SATURATE, ConvertPolicy::WRAP })),
+                                                                   DepthConvertShiftDataset),
+               shape, policy, shift)
+{
+    int fixed_point_position = 0;
+
+    // Create tensors
+    CLTensor src = create_tensor<CLTensor>(shape, DataType::S16, 1, fixed_point_position);
+    CLTensor dst = create_tensor<CLTensor>(shape, DataType::S32, 1, fixed_point_position);
+
+    // Create and Configure function
+    CLDepthConvert depth_convert;
+    depth_convert.configure(&src, &dst, policy, shift);
+
+    // Validate valid region
+    const ValidRegion valid_region = shape_to_valid_region(shape);
+    validate(dst.info()->valid_region(), valid_region);
+
+    // Validate padding
+    const PaddingSize padding = PaddingCalculator(shape.x(), 16).required_padding();
+    validate(src.info()->padding(), padding);
+    validate(dst.info()->padding(), padding);
+}
+
+FIXTURE_DATA_TEST_CASE(RunSmall, CLDepthConvertToS32Fixture<int16_t>, framework::DatasetMode::PRECOMMIT, combine(combine(combine(datasets::SmallShapes(), DepthConvertS16toS32Dataset),
+                                                                                                                 framework::dataset::make("ConvertPolicy", { ConvertPolicy::SATURATE, ConvertPolicy::WRAP })),
+                                                                                                                 DepthConvertShiftDataset))
+{
+    // Validate output
+    validate(CLAccessor(_target), _reference);
+}
+FIXTURE_DATA_TEST_CASE(RunLarge, CLDepthConvertToS32Fixture<int16_t>, framework::DatasetMode::NIGHTLY, combine(combine(combine(datasets::LargeShapes(), DepthConvertS16toS32Dataset),
+                                                                                                                       framework::dataset::make("ConvertPolicy", { ConvertPolicy::SATURATE, ConvertPolicy::WRAP })),
+                                                                                                               DepthConvertShiftDataset))
+{
+    // Validate output
+    validate(CLAccessor(_target), _reference);
+}
+TEST_SUITE_END()
+
+TEST_SUITE(Quantized_to_FP32)
+DATA_TEST_CASE(Configuration, framework::DatasetMode::ALL, combine(combine(combine(framework::dataset::concat(datasets::SmallShapes(), datasets::LargeShapes()), framework::dataset::make("DataType", { DataType::QS8, DataType::QS16 })),
+                                                                           framework::dataset::make("ConvertPolicy", { ConvertPolicy::SATURATE, ConvertPolicy::WRAP })),
+                                                                   DepthConvertFixedPointQuantizedDataset),
+               shape, dt, policy, fixed_point_position)
+{
+    int shift = 0;
+
+    // Create tensors
+    CLTensor src = create_tensor<CLTensor>(shape, dt, 1, fixed_point_position);
+    CLTensor dst = create_tensor<CLTensor>(shape, DataType::F32, 1, fixed_point_position);
+
+    // Create and Configure function
+    CLDepthConvert depth_convert;
+    depth_convert.configure(&src, &dst, policy, shift);
+
+    // Validate valid region
+    const ValidRegion valid_region = shape_to_valid_region(shape);
+    validate(dst.info()->valid_region(), valid_region);
+
+    // Validate padding
+    const PaddingSize padding = PaddingCalculator(shape.x(), 16).required_padding();
+    validate(src.info()->padding(), padding);
+    validate(dst.info()->padding(), padding);
+}
+FIXTURE_DATA_TEST_CASE(RunSmallQS8, CLDepthConvertToFP32FixedPointFixture<int8_t>, framework::DatasetMode::PRECOMMIT, combine(combine(combine(datasets::SmallShapes(), DepthConvertQS8toFP32Dataset),
+                       framework::dataset::make("ConvertPolicy", { ConvertPolicy::SATURATE, ConvertPolicy::WRAP })),
+                       DepthConvertFixedPointQuantizedDataset))
+{
+    // Validate output
+    validate(CLAccessor(_target), _reference);
+}
+FIXTURE_DATA_TEST_CASE(RunSmallQS16, CLDepthConvertToFP32FixedPointFixture<int16_t>, framework::DatasetMode::PRECOMMIT, combine(combine(combine(datasets::SmallShapes(), DepthConvertQS16toFP32Dataset),
+                       framework::dataset::make("ConvertPolicy", { ConvertPolicy::SATURATE, ConvertPolicy::WRAP })),
+                       DepthConvertFixedPointQuantizedDataset))
+{
+    // Validate output
+    validate(CLAccessor(_target), _reference);
+}
+FIXTURE_DATA_TEST_CASE(RunLargeQS8, CLDepthConvertToFP32FixedPointFixture<int8_t>, framework::DatasetMode::NIGHTLY, combine(combine(combine(datasets::LargeShapes(), DepthConvertQS8toFP32Dataset),
+                       framework::dataset::make("ConvertPolicy", { ConvertPolicy::SATURATE, ConvertPolicy::WRAP })),
+                       DepthConvertFixedPointQuantizedDataset))
+{
+    // Validate output
+    validate(CLAccessor(_target), _reference);
+}
+FIXTURE_DATA_TEST_CASE(RunLargeQS16, CLDepthConvertToFP32FixedPointFixture<int16_t>, framework::DatasetMode::NIGHTLY, combine(combine(combine(datasets::LargeShapes(), DepthConvertQS16toFP32Dataset),
+                       framework::dataset::make("ConvertPolicy", { ConvertPolicy::SATURATE, ConvertPolicy::WRAP })),
+                       DepthConvertFixedPointQuantizedDataset))
+{
+    // Validate output
+    validate(CLAccessor(_target), _reference);
+}
+TEST_SUITE_END()
+
+TEST_SUITE(FP32_to_Quantized)
+DATA_TEST_CASE(Configuration, framework::DatasetMode::ALL, combine(combine(combine(framework::dataset::concat(datasets::SmallShapes(), datasets::LargeShapes()), framework::dataset::make("DataType", { DataType::QS8, DataType::QS16 })),
+                                                                           framework::dataset::make("ConvertPolicy", { ConvertPolicy::SATURATE, ConvertPolicy::WRAP })),
+                                                                   DepthConvertFixedPointQuantizedDataset),
+               shape, dt, policy, fixed_point_position)
+{
+    int shift = 0;
+
+    // Create tensors
+    CLTensor src = create_tensor<CLTensor>(shape, DataType::F32, 1, fixed_point_position);
+    CLTensor dst = create_tensor<CLTensor>(shape, dt, 1, fixed_point_position);
+
+    // Create and Configure function
+    CLDepthConvert depth_convert;
+    depth_convert.configure(&src, &dst, policy, shift);
+
+    // Validate valid region
+    const ValidRegion valid_region = shape_to_valid_region(shape);
+    validate(dst.info()->valid_region(), valid_region);
+
+    // Validate padding
+    const PaddingSize padding = PaddingCalculator(shape.x(), 16).required_padding();
+    validate(src.info()->padding(), padding);
+    validate(dst.info()->padding(), padding);
+}
+FIXTURE_DATA_TEST_CASE(RunSmallQS8, CLDepthConvertToQS8FixedPointFixture<float>, framework::DatasetMode::PRECOMMIT, combine(combine(combine(datasets::SmallShapes(), DepthConvertFP32toQS8Dataset),
+                       framework::dataset::make("ConvertPolicy", { ConvertPolicy::SATURATE, ConvertPolicy::WRAP })),
+                       DepthConvertFixedPointQuantizedDataset))
+{
+    // Validate output
+    validate(CLAccessor(_target), _reference);
+}
+FIXTURE_DATA_TEST_CASE(RunSmallQS16, CLDepthConvertToQS16FixedPointFixture<float>, framework::DatasetMode::PRECOMMIT, combine(combine(combine(datasets::SmallShapes(), DepthConvertFP32toQS16Dataset),
+                       framework::dataset::make("ConvertPolicy", { ConvertPolicy::SATURATE, ConvertPolicy::WRAP })),
+                       DepthConvertFixedPointQuantizedDataset))
+{
+    // Validate output
+    validate(CLAccessor(_target), _reference);
+}
+FIXTURE_DATA_TEST_CASE(RunLargeQS8, CLDepthConvertToQS8FixedPointFixture<float>, framework::DatasetMode::NIGHTLY, combine(combine(combine(datasets::LargeShapes(), DepthConvertFP32toQS8Dataset),
+                       framework::dataset::make("ConvertPolicy", { ConvertPolicy::SATURATE, ConvertPolicy::WRAP })),
+                       DepthConvertFixedPointQuantizedDataset))
+{
+    // Validate output
+    validate(CLAccessor(_target), _reference);
+}
+FIXTURE_DATA_TEST_CASE(RunLargeQS16, CLDepthConvertToQS16FixedPointFixture<float>, framework::DatasetMode::NIGHTLY, combine(combine(combine(datasets::LargeShapes(), DepthConvertFP32toQS16Dataset),
+                       framework::dataset::make("ConvertPolicy", { ConvertPolicy::SATURATE, ConvertPolicy::WRAP })),
+                       DepthConvertFixedPointQuantizedDataset))
+{
+    // Validate output
+    validate(CLAccessor(_target), _reference);
+}
+TEST_SUITE_END()
+
+TEST_SUITE_END()
+TEST_SUITE_END()
+} // namespace validation
+} // namespace test
+} // namespace arm_compute
diff --git a/tests/validation/CPP/DepthConvert.cpp b/tests/validation/CPP/DepthConvert.cpp
new file mode 100644
index 0000000000..bb34f67086
--- /dev/null
+++ b/tests/validation/CPP/DepthConvert.cpp
@@ -0,0 +1,158 @@
+/*
+ * Copyright (c) 2017 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 "DepthConvert.h"
+
+#include "tests/validation/FixedPoint.h"
+#include "tests/validation/Helpers.h"
+#include "tests/validation/half.h"
+
+#include "tests/Types.h"
+
+namespace arm_compute
+{
+namespace test
+{
+namespace validation
+{
+namespace reference
+{
+template < typename T1, typename T2, typename std::enable_if < std::is_integral<T1>::value &&std::is_floating_point<T2>::value, int >::type >
+SimpleTensor<T2> depth_convert(const SimpleTensor<T1> &src, DataType dt_out, ConvertPolicy policy, uint32_t shift)
+{
+    ARM_COMPUTE_UNUSED(policy);
+    ARM_COMPUTE_UNUSED(shift);
+
+    using namespace fixed_point_arithmetic;
+    SimpleTensor<T2> result(src.shape(), dt_out);
+
+    const int fixed_point_position = src.fixed_point_position();
+
+    for(int i = 0; i < src.num_elements(); ++i)
+    {
+        result[i] = static_cast<float>(fixed_point<T1>(src[i], fixed_point_position, true));
+    }
+
+    return result;
+}
+
+template < typename T1, typename T2, typename std::enable_if < std::is_floating_point<T1>::value &&std::is_integral<T2>::value, int >::type >
+SimpleTensor<T2> depth_convert(const SimpleTensor<T1> &src, DataType dt_out, ConvertPolicy policy, uint32_t shift)
+{
+    ARM_COMPUTE_UNUSED(policy);
+    ARM_COMPUTE_UNUSED(shift);
+
+    using namespace fixed_point_arithmetic;
+    SimpleTensor<T2> result(src.shape(), dt_out, 1, src.fixed_point_position());
+
+    const int fixed_point_position = result.fixed_point_position();
+
+    for(int i = 0; i < src.num_elements(); ++i)
+    {
+        result[i] = fixed_point<T2>(src[i], fixed_point_position).raw();
+    }
+
+    return result;
+}
+
+template < typename T1, typename T2, typename std::enable_if < std::is_integral<T1>::value &&std::is_integral<T2>::value &&!std::is_same<T1, T2>::value, int >::type >
+SimpleTensor<T2> depth_convert(const SimpleTensor<T1> &src, DataType dt_out, ConvertPolicy policy, uint32_t shift)
+{
+    SimpleTensor<T2> result(src.shape(), dt_out);
+
+    // Up-casting
+    if(src.data_type() <= dt_out)
+    {
+        for(int i = 0; i < src.num_elements(); ++i)
+        {
+            result[i] = src[i] << shift;
+        }
+    }
+    // Down-casting
+    else
+    {
+        for(int i = 0; i < src.num_elements(); ++i)
+        {
+            T1 val    = src[i] >> shift;
+            result[i] = (policy == ConvertPolicy::SATURATE) ? saturate_cast<T2>(val) : static_cast<T2>(val);
+        }
+    }
+    return result;
+}
+
+template < typename T1, typename T2, typename std::enable_if < std::is_integral<T1>::value &&std::is_integral<T2>::value &&std::is_same<T1, T2>::value, int >::type >
+SimpleTensor<T2> depth_convert(const SimpleTensor<T1> &src, DataType dt_out, ConvertPolicy policy, uint32_t shift)
+{
+    ARM_COMPUTE_UNUSED(policy);
+
+    using namespace fixed_point_arithmetic;
+
+    SimpleTensor<T2> result(src.shape(), dt_out);
+
+    bool is_in_place = (&src == &result);
+
+    const int fixed_point_position_in  = src.fixed_point_position();
+    const int fixed_point_position_out = (is_in_place) ? static_cast<int>(shift) : result.fixed_point_position();
+
+    if(!is_in_place || (fixed_point_position_in != fixed_point_position_out))
+    {
+        for(int i = 0; i < src.num_elements(); ++i)
+        {
+            auto x = fixed_point<T2>(src[i], fixed_point_position_in, true);
+            x.resacle(fixed_point_position_out);
+            result[i] = x.raw();
+        }
+    }
+
+    return result;
+}
+
+template < typename T1, typename T2, typename std::enable_if < std::is_floating_point<T1>::value &&is_floating_point<T2>::value, int >::type >
+SimpleTensor<T2> depth_convert(const SimpleTensor<T1> &src, DataType dt_out, ConvertPolicy policy, uint32_t shift)
+{
+    ARM_COMPUTE_UNUSED(policy);
+    ARM_COMPUTE_UNUSED(shift);
+
+    SimpleTensor<T2> result(src.shape(), dt_out);
+
+    for(int i = 0; i < src.num_elements(); ++i)
+    {
+        result[i] = static_cast<T2>(src[i]);
+    }
+}
+
+template SimpleTensor<uint16_t> depth_convert(const SimpleTensor<uint8_t> &src, DataType dt_out, ConvertPolicy policy, uint32_t shift);
+template SimpleTensor<int16_t> depth_convert(const SimpleTensor<uint8_t> &src, DataType dt_out, ConvertPolicy policy, uint32_t shift);
+template SimpleTensor<int32_t> depth_convert(const SimpleTensor<uint8_t> &src, DataType dt_out, ConvertPolicy policy, uint32_t shift);
+template SimpleTensor<uint8_t> depth_convert(const SimpleTensor<uint16_t> &src, DataType dt_out, ConvertPolicy policy, uint32_t shift);
+template SimpleTensor<uint32_t> depth_convert(const SimpleTensor<uint16_t> &src, DataType dt_out, ConvertPolicy policy, uint32_t shift);
+template SimpleTensor<uint8_t> depth_convert(const SimpleTensor<int16_t> &src, DataType dt_out, ConvertPolicy policy, uint32_t shift);
+template SimpleTensor<int32_t> depth_convert(const SimpleTensor<int16_t> &src, DataType dt_out, ConvertPolicy policy, uint32_t shift);
+template SimpleTensor<float> depth_convert(const SimpleTensor<int8_t> &src, DataType dt_out, ConvertPolicy policy, uint32_t shift);
+template SimpleTensor<float> depth_convert(const SimpleTensor<int16_t> &src, DataType dt_out, ConvertPolicy policy, uint32_t shift);
+template SimpleTensor<int8_t> depth_convert(const SimpleTensor<float> &src, DataType dt_out, ConvertPolicy policy, uint32_t shift);
+template SimpleTensor<int16_t> depth_convert(const SimpleTensor<float> &src, DataType dt_out, ConvertPolicy policy, uint32_t shift);
+} // namespace reference
+} // namespace validation
+} // namespace test
+} // namespace arm_compute
diff --git a/tests/validation/CPP/DepthConvert.h b/tests/validation/CPP/DepthConvert.h
new file mode 100644
index 0000000000..1446bfda5b
--- /dev/null
+++ b/tests/validation/CPP/DepthConvert.h
@@ -0,0 +1,56 @@
+/*
+ * Copyright (c) 2017 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.
+ */
+#ifndef __ARM_COMPUTE_TEST_DEPTH_CONVERT_H__
+#define __ARM_COMPUTE_TEST_DEPTH_CONVERT_H__
+
+#include "tests/SimpleTensor.h"
+#include "tests/validation/Helpers.h"
+
+namespace arm_compute
+{
+namespace test
+{
+namespace validation
+{
+namespace reference
+{
+template < typename T1, typename T2, typename std::enable_if < std::is_integral<T1>::value &&std::is_floating_point<T2>::value, int >::type = 0 >
+SimpleTensor<T2> depth_convert(const SimpleTensor<T1> &src, DataType dt_out, ConvertPolicy policy, uint32_t shift);
+
+template < typename T1, typename T2, typename std::enable_if < std::is_floating_point<T1>::value &&std::is_integral<T2>::value, int >::type = 0 >
+SimpleTensor<T2> depth_convert(const SimpleTensor<T1> &src, DataType dt_out, ConvertPolicy policy, uint32_t shift);
+
+template < typename T1, typename T2, typename std::enable_if < std::is_integral<T1>::value &&std::is_integral<T2>::value &&!std::is_same<T1, T2>::value, int >::type = 0 >
+SimpleTensor<T2> depth_convert(const SimpleTensor<T1> &src, DataType dt_out, ConvertPolicy policy, uint32_t shift);
+
+template < typename T1, typename T2, typename std::enable_if < std::is_integral<T1>::value &&std::is_integral<T2>::value &&std::is_same<T1, T2>::value, int >::type = 0 >
+SimpleTensor<T2> depth_convert(const SimpleTensor<T1> &src, DataType dt_out, ConvertPolicy policy, uint32_t shift);
+
+template < typename T1, typename T2, typename std::enable_if < std::is_floating_point<T1>::value &&is_floating_point<T2>::value, int >::type = 0 >
+SimpleTensor<T2> depth_convert(const SimpleTensor<T1> &src, DataType dt_out, ConvertPolicy policy, uint32_t shift);
+} // namespace reference
+} // namespace validation
+} // namespace test
+} // namespace arm_compute
+#endif /* __ARM_COMPUTE_TEST_DEPTH_CONVERT_H__ */
diff --git a/tests/validation/NEON/DepthConvert.cpp b/tests/validation/NEON/DepthConvert.cpp
new file mode 100644
index 0000000000..acc9d0f18b
--- /dev/null
+++ b/tests/validation/NEON/DepthConvert.cpp
@@ -0,0 +1,485 @@
+/*
+ * Copyright (c) 2017 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 CONCLCTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+#include "arm_compute/core/Types.h"
+#include "arm_compute/runtime/NEON/functions/NEDepthConvert.h"
+#include "arm_compute/runtime/Tensor.h"
+#include "arm_compute/runtime/TensorAllocator.h"
+#include "tests/NEON/Accessor.h"
+#include "tests/PaddingCalculator.h"
+#include "tests/datasets/ConvertPolicyDataset.h"
+#include "tests/datasets/ShapeDatasets.h"
+#include "tests/framework/Asserts.h"
+#include "tests/framework/Macros.h"
+#include "tests/framework/datasets/Datasets.h"
+#include "tests/validation/Validation.h"
+#include "tests/validation/fixtures/DepthConvertFixture.h"
+#include "tests/validation/half.h"
+
+namespace arm_compute
+{
+namespace test
+{
+namespace validation
+{
+namespace
+{
+/** Input data sets **/
+const auto DepthConvertU8toU16Dataset             = combine(framework::dataset::make("DataType", DataType::U8), framework::dataset::make("DataType", DataType::U16));
+const auto DepthConvertU8toS16Dataset             = combine(framework::dataset::make("DataType", DataType::U8), framework::dataset::make("DataType", DataType::S16));
+const auto DepthConvertU8toS32Dataset             = combine(framework::dataset::make("DataType", DataType::U8), framework::dataset::make("DataType", DataType::S32));
+const auto DepthConvertU16toU8Dataset             = combine(framework::dataset::make("DataType", DataType::U16), framework::dataset::make("DataType", DataType::U8));
+const auto DepthConvertU16toU32Dataset            = combine(framework::dataset::make("DataType", DataType::U16), framework::dataset::make("DataType", DataType::U32));
+const auto DepthConvertS16toU8Dataset             = combine(framework::dataset::make("DataType", DataType::S16), framework::dataset::make("DataType", DataType::U8));
+const auto DepthConvertS16toS32Dataset            = combine(framework::dataset::make("DataType", DataType::S16), framework::dataset::make("DataType", DataType::S32));
+const auto DepthConvertQS8toFP32Dataset           = combine(framework::dataset::make("DataType", DataType::QS8), framework::dataset::make("DataType", DataType::F32));
+const auto DepthConvertQS16toFP32Dataset          = combine(framework::dataset::make("DataType", DataType::QS16), framework::dataset::make("DataType", DataType::F32));
+const auto DepthConvertFP32toQS8Dataset           = combine(framework::dataset::make("DataType", DataType::F32), framework::dataset::make("DataType", DataType::QS8));
+const auto DepthConvertFP32toQS16Dataset          = combine(framework::dataset::make("DataType", DataType::F32), framework::dataset::make("DataType", DataType::QS16));
+const auto DepthConvertShiftDataset               = framework::dataset::make("Shift", 0, 7);
+const auto DepthConvertFixedPointQuantizedDataset = framework::dataset::make("FractionalBits", 1, 7);
+} // namespace
+
+TEST_SUITE(NEON)
+TEST_SUITE(DepthConvert)
+template <typename T>
+using NEDepthConvertToU16Fixture = DepthConvertValidationFixture<Tensor, Accessor, NEDepthConvert, T, uint16_t>;
+template <typename T>
+using NEDepthConvertToS16Fixture = DepthConvertValidationFixture<Tensor, Accessor, NEDepthConvert, T, int16_t>;
+template <typename T>
+using NEDepthConvertToS32Fixture = DepthConvertValidationFixture<Tensor, Accessor, NEDepthConvert, T, int32_t>;
+template <typename T>
+using NEDepthConvertToU8Fixture = DepthConvertValidationFixture<Tensor, Accessor, NEDepthConvert, T, uint8_t>;
+template <typename T>
+using NEDepthConvertToU32Fixture = DepthConvertValidationFixture<Tensor, Accessor, NEDepthConvert, T, uint32_t>;
+template <typename T>
+using NEDepthConvertToFP32FixedPointFixture = DepthConvertValidationFractionalBitsFixture<Tensor, Accessor, NEDepthConvert, T, float>;
+template <typename T>
+using NEDepthConvertToQS8FixedPointFixture = DepthConvertValidationFractionalBitsFixture<Tensor, Accessor, NEDepthConvert, T, int8_t>;
+template <typename T>
+using NEDepthConvertToQS16FixedPointFixture = DepthConvertValidationFractionalBitsFixture<Tensor, Accessor, NEDepthConvert, T, int16_t>;
+
+TEST_SUITE(U8_to_U16)
+DATA_TEST_CASE(Configuration, framework::DatasetMode::ALL, combine(combine(framework::dataset::concat(datasets::SmallShapes(), datasets::LargeShapes()), framework::dataset::make("ConvertPolicy", { ConvertPolicy::SATURATE, ConvertPolicy::WRAP })),
+                                                                   DepthConvertShiftDataset),
+               shape, policy, shift)
+{
+    int fixed_point_position = 0;
+
+    // Create tensors
+    Tensor src = create_tensor<Tensor>(shape, DataType::U8, 1, fixed_point_position);
+    Tensor dst = create_tensor<Tensor>(shape, DataType::U16, 1, fixed_point_position);
+
+    // Create and Configure function
+    NEDepthConvert depth_convert;
+    depth_convert.configure(&src, &dst, policy, shift);
+
+    // Validate valid region
+    const ValidRegion valid_region = shape_to_valid_region(shape);
+    validate(dst.info()->valid_region(), valid_region);
+
+    // Validate padding
+    const PaddingSize padding = PaddingCalculator(shape.x(), 16).required_padding();
+    validate(src.info()->padding(), padding);
+    validate(dst.info()->padding(), padding);
+}
+
+FIXTURE_DATA_TEST_CASE(RunSmall, NEDepthConvertToU16Fixture<uint8_t>, framework::DatasetMode::PRECOMMIT, combine(combine(combine(datasets::SmallShapes(), DepthConvertU8toU16Dataset),
+                                                                                                                 framework::dataset::make("ConvertPolicy", { ConvertPolicy::SATURATE, ConvertPolicy::WRAP })),
+                                                                                                                 DepthConvertShiftDataset))
+{
+    // Validate output
+    validate(Accessor(_target), _reference);
+}
+
+FIXTURE_DATA_TEST_CASE(RunLarge, NEDepthConvertToU16Fixture<uint8_t>, framework::DatasetMode::NIGHTLY, combine(combine(combine(datasets::LargeShapes(), DepthConvertU8toU16Dataset),
+                                                                                                                       framework::dataset::make("ConvertPolicy", { ConvertPolicy::SATURATE, ConvertPolicy::WRAP })),
+                                                                                                               DepthConvertShiftDataset))
+{
+    // Validate output
+    validate(Accessor(_target), _reference);
+}
+TEST_SUITE_END()
+
+TEST_SUITE(U8_to_S16)
+DATA_TEST_CASE(Configuration, framework::DatasetMode::ALL, combine(combine(framework::dataset::concat(datasets::SmallShapes(), datasets::LargeShapes()), framework::dataset::make("ConvertPolicy", { ConvertPolicy::SATURATE, ConvertPolicy::WRAP })),
+                                                                   DepthConvertShiftDataset),
+               shape, policy, shift)
+{
+    int fixed_point_position = 0;
+
+    // Create tensors
+    Tensor src = create_tensor<Tensor>(shape, DataType::U8, 1, fixed_point_position);
+    Tensor dst = create_tensor<Tensor>(shape, DataType::S16, 1, fixed_point_position);
+
+    // Create and Configure function
+    NEDepthConvert depth_convert;
+    depth_convert.configure(&src, &dst, policy, shift);
+
+    // Validate valid region
+    const ValidRegion valid_region = shape_to_valid_region(shape);
+    validate(dst.info()->valid_region(), valid_region);
+
+    // Validate padding
+    const PaddingSize padding = PaddingCalculator(shape.x(), 16).required_padding();
+    validate(src.info()->padding(), padding);
+    validate(dst.info()->padding(), padding);
+}
+
+FIXTURE_DATA_TEST_CASE(RunSmall, NEDepthConvertToS16Fixture<uint8_t>, framework::DatasetMode::PRECOMMIT, combine(combine(combine(datasets::SmallShapes(), DepthConvertU8toS16Dataset),
+                                                                                                                 framework::dataset::make("ConvertPolicy", { ConvertPolicy::SATURATE, ConvertPolicy::WRAP })),
+                                                                                                                 DepthConvertShiftDataset))
+{
+    // Validate output
+    validate(Accessor(_target), _reference);
+}
+
+FIXTURE_DATA_TEST_CASE(RunLarge, NEDepthConvertToS16Fixture<uint8_t>, framework::DatasetMode::NIGHTLY, combine(combine(combine(datasets::LargeShapes(), DepthConvertU8toS16Dataset),
+                                                                                                                       framework::dataset::make("ConvertPolicy", { ConvertPolicy::SATURATE, ConvertPolicy::WRAP })),
+                                                                                                               DepthConvertShiftDataset))
+{
+    // Validate output
+    validate(Accessor(_target), _reference);
+}
+TEST_SUITE_END()
+TEST_SUITE(U8_to_S32)
+DATA_TEST_CASE(Configuration, framework::DatasetMode::ALL, combine(combine(framework::dataset::concat(datasets::SmallShapes(), datasets::LargeShapes()), framework::dataset::make("ConvertPolicy", { ConvertPolicy::SATURATE, ConvertPolicy::WRAP })),
+                                                                   DepthConvertShiftDataset),
+               shape, policy, shift)
+{
+    int fixed_point_position = 0;
+
+    // Create tensors
+    Tensor src = create_tensor<Tensor>(shape, DataType::U8, 1, fixed_point_position);
+    Tensor dst = create_tensor<Tensor>(shape, DataType::S32, 1, fixed_point_position);
+
+    // Create and Configure function
+    NEDepthConvert depth_convert;
+    depth_convert.configure(&src, &dst, policy, shift);
+
+    // Validate valid region
+    const ValidRegion valid_region = shape_to_valid_region(shape);
+    validate(dst.info()->valid_region(), valid_region);
+
+    // Validate padding
+    const PaddingSize padding = PaddingCalculator(shape.x(), 16).required_padding();
+    validate(src.info()->padding(), padding);
+    validate(dst.info()->padding(), padding);
+}
+
+FIXTURE_DATA_TEST_CASE(RunSmall, NEDepthConvertToS32Fixture<uint8_t>, framework::DatasetMode::PRECOMMIT, combine(combine(combine(datasets::SmallShapes(), DepthConvertU8toS32Dataset),
+                                                                                                                 framework::dataset::make("ConvertPolicy", { ConvertPolicy::SATURATE, ConvertPolicy::WRAP })),
+                                                                                                                 DepthConvertShiftDataset))
+{
+    // Validate output
+    validate(Accessor(_target), _reference);
+}
+
+FIXTURE_DATA_TEST_CASE(RunLarge, NEDepthConvertToS32Fixture<uint8_t>, framework::DatasetMode::NIGHTLY, combine(combine(combine(datasets::LargeShapes(), DepthConvertU8toS32Dataset),
+                                                                                                                       framework::dataset::make("ConvertPolicy", { ConvertPolicy::SATURATE, ConvertPolicy::WRAP })),
+                                                                                                               DepthConvertShiftDataset))
+{
+    // Validate output
+    validate(Accessor(_target), _reference);
+}
+TEST_SUITE_END()
+
+TEST_SUITE(U16_to_U8)
+DATA_TEST_CASE(Configuration, framework::DatasetMode::ALL, combine(combine(framework::dataset::concat(datasets::SmallShapes(), datasets::LargeShapes()), framework::dataset::make("ConvertPolicy", { ConvertPolicy::SATURATE, ConvertPolicy::WRAP })),
+                                                                   DepthConvertShiftDataset),
+               shape, policy, shift)
+{
+    int fixed_point_position = 0;
+
+    // Create tensors
+    Tensor src = create_tensor<Tensor>(shape, DataType::U16, 1, fixed_point_position);
+    Tensor dst = create_tensor<Tensor>(shape, DataType::U8, 1, fixed_point_position);
+
+    // Create and Configure function
+    NEDepthConvert depth_convert;
+    depth_convert.configure(&src, &dst, policy, shift);
+
+    // Validate valid region
+    const ValidRegion valid_region = shape_to_valid_region(shape);
+    validate(dst.info()->valid_region(), valid_region);
+
+    // Validate padding
+    const PaddingSize padding = PaddingCalculator(shape.x(), 16).required_padding();
+    validate(src.info()->padding(), padding);
+    validate(dst.info()->padding(), padding);
+}
+
+FIXTURE_DATA_TEST_CASE(RunSmall, NEDepthConvertToU8Fixture<uint16_t>, framework::DatasetMode::PRECOMMIT, combine(combine(combine(datasets::SmallShapes(), DepthConvertU16toU8Dataset),
+                                                                                                                 framework::dataset::make("ConvertPolicy", { ConvertPolicy::SATURATE, ConvertPolicy::WRAP })),
+                                                                                                                 DepthConvertShiftDataset))
+{
+    // Validate output
+    validate(Accessor(_target), _reference);
+}
+FIXTURE_DATA_TEST_CASE(RunLarge, NEDepthConvertToU8Fixture<uint16_t>, framework::DatasetMode::NIGHTLY, combine(combine(combine(datasets::LargeShapes(), DepthConvertU16toU8Dataset),
+                                                                                                                       framework::dataset::make("ConvertPolicy", { ConvertPolicy::SATURATE, ConvertPolicy::WRAP })),
+                                                                                                               DepthConvertShiftDataset))
+{
+    // Validate output
+    validate(Accessor(_target), _reference);
+}
+TEST_SUITE_END()
+
+TEST_SUITE(U16_to_U32)
+DATA_TEST_CASE(Configuration, framework::DatasetMode::ALL, combine(combine(framework::dataset::concat(datasets::SmallShapes(), datasets::LargeShapes()), framework::dataset::make("ConvertPolicy", { ConvertPolicy::SATURATE, ConvertPolicy::WRAP })),
+                                                                   DepthConvertShiftDataset),
+               shape, policy, shift)
+{
+    int fixed_point_position = 0;
+
+    // Create tensors
+    Tensor src = create_tensor<Tensor>(shape, DataType::U16, 1, fixed_point_position);
+    Tensor dst = create_tensor<Tensor>(shape, DataType::U32, 1, fixed_point_position);
+
+    // Create and Configure function
+    NEDepthConvert depth_convert;
+    depth_convert.configure(&src, &dst, policy, shift);
+
+    // Validate valid region
+    const ValidRegion valid_region = shape_to_valid_region(shape);
+    validate(dst.info()->valid_region(), valid_region);
+
+    // Validate padding
+    const PaddingSize padding = PaddingCalculator(shape.x(), 16).required_padding();
+    validate(src.info()->padding(), padding);
+    validate(dst.info()->padding(), padding);
+}
+
+FIXTURE_DATA_TEST_CASE(RunSmall, NEDepthConvertToU32Fixture<uint16_t>, framework::DatasetMode::PRECOMMIT, combine(combine(combine(datasets::SmallShapes(), DepthConvertU16toU32Dataset),
+                                                                                                                  framework::dataset::make("ConvertPolicy", { ConvertPolicy::SATURATE, ConvertPolicy::WRAP })),
+                                                                                                                  DepthConvertShiftDataset))
+{
+    // Validate output
+    validate(Accessor(_target), _reference);
+}
+FIXTURE_DATA_TEST_CASE(RunLarge, NEDepthConvertToU32Fixture<uint16_t>, framework::DatasetMode::NIGHTLY, combine(combine(combine(datasets::LargeShapes(), DepthConvertU16toU32Dataset),
+                                                                                                                        framework::dataset::make("ConvertPolicy", { ConvertPolicy::SATURATE, ConvertPolicy::WRAP })),
+                                                                                                                DepthConvertShiftDataset))
+{
+    // Validate output
+    validate(Accessor(_target), _reference);
+}
+TEST_SUITE_END()
+
+TEST_SUITE(S16_to_U8)
+DATA_TEST_CASE(Configuration, framework::DatasetMode::ALL, combine(combine(framework::dataset::concat(datasets::SmallShapes(), datasets::LargeShapes()), framework::dataset::make("ConvertPolicy", { ConvertPolicy::SATURATE, ConvertPolicy::WRAP })),
+                                                                   DepthConvertShiftDataset),
+               shape, policy, shift)
+{
+    int fixed_point_position = 0;
+
+    // Create tensors
+    Tensor src = create_tensor<Tensor>(shape, DataType::S16, 1, fixed_point_position);
+    Tensor dst = create_tensor<Tensor>(shape, DataType::U8, 1, fixed_point_position);
+
+    // Create and Configure function
+    NEDepthConvert depth_convert;
+    depth_convert.configure(&src, &dst, policy, shift);
+
+    // Validate valid region
+    const ValidRegion valid_region = shape_to_valid_region(shape);
+    validate(dst.info()->valid_region(), valid_region);
+
+    // Validate padding
+    const PaddingSize padding = PaddingCalculator(shape.x(), 16).required_padding();
+    validate(src.info()->padding(), padding);
+    validate(dst.info()->padding(), padding);
+}
+
+FIXTURE_DATA_TEST_CASE(RunSmall, NEDepthConvertToU8Fixture<int16_t>, framework::DatasetMode::PRECOMMIT, combine(combine(combine(datasets::SmallShapes(), DepthConvertS16toU8Dataset),
+                                                                                                                        framework::dataset::make("ConvertPolicy", { ConvertPolicy::SATURATE, ConvertPolicy::WRAP })),
+                                                                                                                DepthConvertShiftDataset))
+{
+    // Validate output
+    validate(Accessor(_target), _reference);
+}
+FIXTURE_DATA_TEST_CASE(RunLarge, NEDepthConvertToU8Fixture<int16_t>, framework::DatasetMode::NIGHTLY, combine(combine(combine(datasets::LargeShapes(), DepthConvertS16toU8Dataset),
+                                                                                                                      framework::dataset::make("ConvertPolicy", { ConvertPolicy::SATURATE, ConvertPolicy::WRAP })),
+                                                                                                              DepthConvertShiftDataset))
+{
+    // Validate output
+    validate(Accessor(_target), _reference);
+}
+TEST_SUITE_END()
+
+TEST_SUITE(S16_to_S32)
+DATA_TEST_CASE(Configuration, framework::DatasetMode::ALL, combine(combine(framework::dataset::concat(datasets::SmallShapes(), datasets::LargeShapes()), framework::dataset::make("ConvertPolicy", { ConvertPolicy::SATURATE, ConvertPolicy::WRAP })),
+                                                                   DepthConvertShiftDataset),
+               shape, policy, shift)
+{
+    int fixed_point_position = 0;
+
+    // Create tensors
+    Tensor src = create_tensor<Tensor>(shape, DataType::S16, 1, fixed_point_position);
+    Tensor dst = create_tensor<Tensor>(shape, DataType::S32, 1, fixed_point_position);
+
+    // Create and Configure function
+    NEDepthConvert depth_convert;
+    depth_convert.configure(&src, &dst, policy, shift);
+
+    // Validate valid region
+    const ValidRegion valid_region = shape_to_valid_region(shape);
+    validate(dst.info()->valid_region(), valid_region);
+
+    // Validate padding
+    const PaddingSize padding = PaddingCalculator(shape.x(), 16).required_padding();
+    validate(src.info()->padding(), padding);
+    validate(dst.info()->padding(), padding);
+}
+
+FIXTURE_DATA_TEST_CASE(RunSmall, NEDepthConvertToS32Fixture<int16_t>, framework::DatasetMode::PRECOMMIT, combine(combine(combine(datasets::SmallShapes(), DepthConvertS16toS32Dataset),
+                                                                                                                 framework::dataset::make("ConvertPolicy", { ConvertPolicy::SATURATE, ConvertPolicy::WRAP })),
+                                                                                                                 DepthConvertShiftDataset))
+{
+    // Validate output
+    validate(Accessor(_target), _reference);
+}
+FIXTURE_DATA_TEST_CASE(RunLarge, NEDepthConvertToS32Fixture<int16_t>, framework::DatasetMode::NIGHTLY, combine(combine(combine(datasets::LargeShapes(), DepthConvertS16toS32Dataset),
+                                                                                                                       framework::dataset::make("ConvertPolicy", { ConvertPolicy::SATURATE, ConvertPolicy::WRAP })),
+                                                                                                               DepthConvertShiftDataset))
+{
+    // Validate output
+    validate(Accessor(_target), _reference);
+}
+TEST_SUITE_END()
+
+TEST_SUITE(Quantized_to_FP32)
+DATA_TEST_CASE(Configuration, framework::DatasetMode::ALL, combine(combine(combine(framework::dataset::concat(datasets::SmallShapes(), datasets::LargeShapes()), framework::dataset::make("DataType", { DataType::QS8, DataType::QS16 })),
+                                                                           framework::dataset::make("ConvertPolicy", { ConvertPolicy::SATURATE, ConvertPolicy::WRAP })),
+                                                                   DepthConvertFixedPointQuantizedDataset),
+               shape, dt, policy, fixed_point_position)
+{
+    int shift = 0;
+
+    // Create tensors
+    Tensor src = create_tensor<Tensor>(shape, dt, 1, fixed_point_position);
+    Tensor dst = create_tensor<Tensor>(shape, DataType::F32, 1, fixed_point_position);
+
+    // Create and Configure function
+    NEDepthConvert depth_convert;
+    depth_convert.configure(&src, &dst, policy, shift);
+
+    // Validate valid region
+    const ValidRegion valid_region = shape_to_valid_region(shape);
+    validate(dst.info()->valid_region(), valid_region);
+
+    // Validate padding
+    const PaddingSize padding = PaddingCalculator(shape.x(), 16).required_padding();
+    validate(src.info()->padding(), padding);
+    validate(dst.info()->padding(), padding);
+}
+FIXTURE_DATA_TEST_CASE(RunSmallQS8, NEDepthConvertToFP32FixedPointFixture<int8_t>, framework::DatasetMode::PRECOMMIT, combine(combine(combine(datasets::SmallShapes(), DepthConvertQS8toFP32Dataset),
+                       framework::dataset::make("ConvertPolicy", { ConvertPolicy::SATURATE, ConvertPolicy::WRAP })),
+                       DepthConvertFixedPointQuantizedDataset))
+{
+    // Validate output
+    validate(Accessor(_target), _reference);
+}
+FIXTURE_DATA_TEST_CASE(RunSmallQS16, NEDepthConvertToFP32FixedPointFixture<int16_t>, framework::DatasetMode::PRECOMMIT, combine(combine(combine(datasets::SmallShapes(), DepthConvertQS16toFP32Dataset),
+                       framework::dataset::make("ConvertPolicy", { ConvertPolicy::SATURATE, ConvertPolicy::WRAP })),
+                       DepthConvertFixedPointQuantizedDataset))
+{
+    // Validate output
+    validate(Accessor(_target), _reference);
+}
+FIXTURE_DATA_TEST_CASE(RunLargeQS8, NEDepthConvertToFP32FixedPointFixture<int8_t>, framework::DatasetMode::NIGHTLY, combine(combine(combine(datasets::LargeShapes(), DepthConvertQS8toFP32Dataset),
+                       framework::dataset::make("ConvertPolicy", { ConvertPolicy::SATURATE, ConvertPolicy::WRAP })),
+                       DepthConvertFixedPointQuantizedDataset))
+{
+    // Validate output
+    validate(Accessor(_target), _reference);
+}
+FIXTURE_DATA_TEST_CASE(RunLargeQS16, NEDepthConvertToFP32FixedPointFixture<int16_t>, framework::DatasetMode::NIGHTLY, combine(combine(combine(datasets::LargeShapes(), DepthConvertQS16toFP32Dataset),
+                       framework::dataset::make("ConvertPolicy", { ConvertPolicy::SATURATE, ConvertPolicy::WRAP })),
+                       DepthConvertFixedPointQuantizedDataset))
+{
+    // Validate output
+    validate(Accessor(_target), _reference);
+}
+TEST_SUITE_END()
+
+TEST_SUITE(FP32_to_Quantized)
+DATA_TEST_CASE(Configuration, framework::DatasetMode::ALL, combine(combine(combine(framework::dataset::concat(datasets::SmallShapes(), datasets::LargeShapes()), framework::dataset::make("DataType", { DataType::QS8, DataType::QS16 })),
+                                                                           framework::dataset::make("ConvertPolicy", { ConvertPolicy::SATURATE, ConvertPolicy::WRAP })),
+                                                                   DepthConvertFixedPointQuantizedDataset),
+               shape, dt, policy, fixed_point_position)
+{
+    int shift = 0;
+
+    // Create tensors
+    Tensor src = create_tensor<Tensor>(shape, DataType::F32, 1, fixed_point_position);
+    Tensor dst = create_tensor<Tensor>(shape, dt, 1, fixed_point_position);
+
+    // Create and Configure function
+    NEDepthConvert depth_convert;
+    depth_convert.configure(&src, &dst, policy, shift);
+
+    // Validate valid region
+    const ValidRegion valid_region = shape_to_valid_region(shape);
+    validate(dst.info()->valid_region(), valid_region);
+
+    // Validate padding
+    const PaddingSize padding = PaddingCalculator(shape.x(), 16).required_padding();
+    validate(src.info()->padding(), padding);
+    validate(dst.info()->padding(), padding);
+}
+FIXTURE_DATA_TEST_CASE(RunSmallQS8, NEDepthConvertToQS8FixedPointFixture<float>, framework::DatasetMode::PRECOMMIT, combine(combine(combine(datasets::SmallShapes(), DepthConvertFP32toQS8Dataset),
+                       framework::dataset::make("ConvertPolicy", { ConvertPolicy::SATURATE, ConvertPolicy::WRAP })),
+                       DepthConvertFixedPointQuantizedDataset))
+{
+    // Validate output
+    validate(Accessor(_target), _reference);
+}
+FIXTURE_DATA_TEST_CASE(RunSmallQS16, NEDepthConvertToQS16FixedPointFixture<float>, framework::DatasetMode::PRECOMMIT, combine(combine(combine(datasets::SmallShapes(), DepthConvertFP32toQS16Dataset),
+                       framework::dataset::make("ConvertPolicy", { ConvertPolicy::SATURATE, ConvertPolicy::WRAP })),
+                       DepthConvertFixedPointQuantizedDataset))
+{
+    // Validate output
+    validate(Accessor(_target), _reference);
+}
+FIXTURE_DATA_TEST_CASE(RunLargeQS8, NEDepthConvertToQS8FixedPointFixture<float>, framework::DatasetMode::NIGHTLY, combine(combine(combine(datasets::LargeShapes(), DepthConvertFP32toQS8Dataset),
+                       framework::dataset::make("ConvertPolicy", { ConvertPolicy::SATURATE, ConvertPolicy::WRAP })),
+                       DepthConvertFixedPointQuantizedDataset))
+{
+    // Validate output
+    validate(Accessor(_target), _reference);
+}
+FIXTURE_DATA_TEST_CASE(RunLargeQS16, NEDepthConvertToQS16FixedPointFixture<float>, framework::DatasetMode::NIGHTLY, combine(combine(combine(datasets::LargeShapes(), DepthConvertFP32toQS16Dataset),
+                       framework::dataset::make("ConvertPolicy", { ConvertPolicy::SATURATE, ConvertPolicy::WRAP })),
+                       DepthConvertFixedPointQuantizedDataset))
+{
+    // Validate output
+    validate(Accessor(_target), _reference);
+}
+TEST_SUITE_END()
+
+TEST_SUITE_END()
+TEST_SUITE_END()
+} // namespace validation
+} // namespace test
+} // namespace arm_compute
diff --git a/tests/validation/fixtures/DepthConvertFixture.h b/tests/validation/fixtures/DepthConvertFixture.h
new file mode 100644
index 0000000000..b132a9341d
--- /dev/null
+++ b/tests/validation/fixtures/DepthConvertFixture.h
@@ -0,0 +1,131 @@
+/*
+ * Copyright (c) 2017 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.
+ */
+#ifndef ARM_COMPUTE_TEST_DEPTH_CONVERT_FIXTURE
+#define ARM_COMPUTE_TEST_DEPTH_CONVERT_FIXTURE
+
+#include "arm_compute/core/TensorShape.h"
+#include "arm_compute/core/Types.h"
+#include "tests/AssetsLibrary.h"
+#include "tests/Globals.h"
+#include "tests/IAccessor.h"
+#include "tests/framework/Asserts.h"
+#include "tests/framework/Fixture.h"
+#include "tests/validation/CPP/DepthConvert.h"
+#include "tests/validation/Helpers.h"
+
+namespace arm_compute
+{
+namespace test
+{
+namespace validation
+{
+template <typename TensorType, typename AccessorType, typename FunctionType, typename T1, typename T2>
+class DepthConvertValidationFixedPointFixture : public framework::Fixture
+{
+public:
+    template <typename...>
+    void setup(TensorShape shape, DataType dt_in, DataType dt_out, ConvertPolicy policy, uint32_t shift, uint32_t fractional_bits)
+    {
+        _shift           = shift;
+        _fractional_bits = fractional_bits;
+        _target          = compute_target(shape, dt_in, dt_out, policy, shift, fractional_bits);
+        _reference       = compute_reference(shape, dt_in, dt_out, policy, shift, fractional_bits);
+    }
+
+protected:
+    template <typename U>
+    void fill(U &&tensor, int i)
+    {
+        library->fill_tensor_uniform(tensor, i);
+    }
+
+    TensorType compute_target(const TensorShape &shape, DataType dt_in, DataType dt_out, ConvertPolicy policy, uint32_t shift, uint32_t fixed_point_position)
+    {
+        // Create tensors
+        TensorType src = create_tensor<TensorType>(shape, dt_in, 1, static_cast<int>(fixed_point_position));
+        TensorType dst = create_tensor<TensorType>(shape, dt_out, 1, static_cast<int>(fixed_point_position));
+
+        // Create and configure function
+        FunctionType depth_convert;
+        depth_convert.configure(&src, &dst, policy, shift);
+
+        ARM_COMPUTE_EXPECT(src.info()->is_resizable(), framework::LogLevel::ERRORS);
+        ARM_COMPUTE_EXPECT(dst.info()->is_resizable(), framework::LogLevel::ERRORS);
+
+        // Allocate tensors
+        src.allocator()->allocate();
+        dst.allocator()->allocate();
+
+        ARM_COMPUTE_EXPECT(!src.info()->is_resizable(), framework::LogLevel::ERRORS);
+        ARM_COMPUTE_EXPECT(!dst.info()->is_resizable(), framework::LogLevel::ERRORS);
+
+        // Fill tensors
+        fill(AccessorType(src), 0);
+
+        // Compute function
+        depth_convert.run();
+
+        return dst;
+    }
+
+    SimpleTensor<T2> compute_reference(const TensorShape &shape, DataType dt_in, DataType dt_out, ConvertPolicy policy, uint32_t shift, uint32_t fixed_point_position)
+    {
+        // Create reference
+        SimpleTensor<T1> src{ shape, dt_in, 1, static_cast<int>(fixed_point_position) };
+
+        // Fill reference
+        fill(src, 0);
+
+        return reference::depth_convert<T1, T2>(src, dt_out, policy, shift);
+    }
+
+    TensorType       _target{};
+    SimpleTensor<T2> _reference{};
+    int              _fractional_bits{};
+    int              _shift{};
+};
+template <typename TensorType, typename AccessorType, typename FunctionType, typename T1, typename T2>
+class DepthConvertValidationFixture : public DepthConvertValidationFixedPointFixture<TensorType, AccessorType, FunctionType, T1, T2>
+{
+public:
+    template <typename...>
+    void setup(TensorShape shape, DataType dt_in, DataType dt_out, ConvertPolicy policy, uint32_t shift)
+    {
+        DepthConvertValidationFixedPointFixture<TensorType, AccessorType, FunctionType, T1, T2>::setup(shape, dt_in, dt_out, policy, shift, 0);
+    }
+};
+template <typename TensorType, typename AccessorType, typename FunctionType, typename T1, typename T2>
+class DepthConvertValidationFractionalBitsFixture : public DepthConvertValidationFixedPointFixture<TensorType, AccessorType, FunctionType, T1, T2>
+{
+public:
+    template <typename...>
+    void setup(TensorShape shape, DataType dt_in, DataType dt_out, ConvertPolicy policy, uint32_t fractional_bits)
+    {
+        DepthConvertValidationFixedPointFixture<TensorType, AccessorType, FunctionType, T1, T2>::setup(shape, dt_in, dt_out, policy, 0, fractional_bits);
+    }
+};
+} // namespace validation
+} // namespace test
+} // namespace arm_compute
+#endif /* ARM_COMPUTE_TEST_DEPTH_CONVERT_FIXTURE */