diff --git a/tests/datasets/SpaceToBatchDataset.h b/tests/datasets/SpaceToBatchDataset.h
new file mode 100644
index 0000000..0ab06f9
--- /dev/null
+++ b/tests/datasets/SpaceToBatchDataset.h
@@ -0,0 +1,142 @@
+/*
+ * Copyright (c) 2018 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_SPACE_TO_BATCH_LAYER_DATASET
+#define ARM_COMPUTE_TEST_SPACE_TO_BATCH_LAYER_DATASET
+
+#include "utils/TypePrinter.h"
+
+#include "arm_compute/core/TensorShape.h"
+#include "arm_compute/core/Types.h"
+
+namespace arm_compute
+{
+namespace test
+{
+namespace datasets
+{
+class SpaceToBatchLayerDataset
+{
+public:
+    using type = std::tuple<TensorShape, TensorShape, TensorShape, TensorShape>;
+
+    struct iterator
+    {
+        iterator(std::vector<TensorShape>::const_iterator src_it,
+                 std::vector<TensorShape>::const_iterator block_shape_it,
+                 std::vector<TensorShape>::const_iterator paddings_shape_it,
+                 std::vector<TensorShape>::const_iterator dst_it)
+            : _src_it{ std::move(src_it) },
+              _block_shape_it{ std::move(block_shape_it) },
+              _paddings_shape_it{ std::move(paddings_shape_it) },
+              _dst_it{ std::move(dst_it) }
+        {
+        }
+
+        std::string description() const
+        {
+            std::stringstream description;
+            description << "In=" << *_src_it << ":";
+            description << "BlockShape=" << *_block_shape_it << ":";
+            description << "PaddingsShape=" << *_paddings_shape_it << ":";
+            description << "Out=" << *_dst_it;
+            return description.str();
+        }
+
+        SpaceToBatchLayerDataset::type operator*() const
+        {
+            return std::make_tuple(*_src_it, *_block_shape_it, *_paddings_shape_it, *_dst_it);
+        }
+
+        iterator &operator++()
+        {
+            ++_src_it;
+            ++_block_shape_it;
+            ++_paddings_shape_it;
+            ++_dst_it;
+
+            return *this;
+        }
+
+    private:
+        std::vector<TensorShape>::const_iterator _src_it;
+        std::vector<TensorShape>::const_iterator _block_shape_it;
+        std::vector<TensorShape>::const_iterator _paddings_shape_it;
+        std::vector<TensorShape>::const_iterator _dst_it;
+    };
+
+    iterator begin() const
+    {
+        return iterator(_src_shapes.begin(), _block_shape_shapes.begin(), _padding_shapes.begin(), _dst_shapes.begin());
+    }
+
+    int size() const
+    {
+        return std::min(_src_shapes.size(), std::min(_block_shape_shapes.size(), std::min(_padding_shapes.size(), _dst_shapes.size())));
+    }
+
+    void add_config(TensorShape src, TensorShape block_shape, TensorShape padding_shapes, TensorShape dst)
+    {
+        _src_shapes.emplace_back(std::move(src));
+        _block_shape_shapes.emplace_back(std::move(block_shape));
+        _padding_shapes.emplace_back(std::move(padding_shapes));
+        _dst_shapes.emplace_back(std::move(dst));
+    }
+
+protected:
+    SpaceToBatchLayerDataset()                            = default;
+    SpaceToBatchLayerDataset(SpaceToBatchLayerDataset &&) = default;
+
+private:
+    std::vector<TensorShape> _src_shapes{};
+    std::vector<TensorShape> _block_shape_shapes{};
+    std::vector<TensorShape> _padding_shapes{};
+    std::vector<TensorShape> _dst_shapes{};
+};
+
+class SmallSpaceToBatchLayerDataset final : public SpaceToBatchLayerDataset
+{
+public:
+    SmallSpaceToBatchLayerDataset()
+    {
+        add_config(TensorShape(2U, 2U, 1U, 1U), TensorShape(2U), TensorShape(2U, 4U), TensorShape(1U, 1U, 1U, 4U));
+        add_config(TensorShape(6U, 2U, 1U, 1U), TensorShape(2U), TensorShape(2U, 4U), TensorShape(3U, 1U, 1U, 4U));
+        add_config(TensorShape(2U, 4U, 2U, 1U), TensorShape(2U), TensorShape(2U, 4U), TensorShape(1U, 2U, 2U, 4U));
+        add_config(TensorShape(2U, 6U, 1U, 2U), TensorShape(2U), TensorShape(2U, 4U), TensorShape(1U, 3U, 1U, 8U));
+        add_config(TensorShape(6U, 8U, 1U, 1U), TensorShape(2U), TensorShape(2U, 4U), TensorShape(3U, 4U, 1U, 4U));
+        add_config(TensorShape(6U, 8U, 15U, 5U), TensorShape(2U), TensorShape(2U, 4U), TensorShape(3U, 4U, 15U, 20U));
+    }
+};
+class LargeSpaceToBatchLayerDataset final : public SpaceToBatchLayerDataset
+{
+public:
+    LargeSpaceToBatchLayerDataset()
+    {
+        add_config(TensorShape(128U, 64U, 2U, 1U), TensorShape(2U), TensorShape(2U, 4U), TensorShape(64U, 32U, 2U, 4U));
+        add_config(TensorShape(512U, 64U, 2U, 1U), TensorShape(2U), TensorShape(2U, 4U), TensorShape(128U, 16U, 2U, 16U));
+    }
+};
+} // namespace datasets
+} // namespace test
+} // namespace arm_compute
+#endif /* ARM_COMPUTE_TEST_SPACE_TO_BATCH_LAYER_DATASET */
diff --git a/tests/validation/CL/SpaceToBatchLayer.cpp b/tests/validation/CL/SpaceToBatchLayer.cpp
new file mode 100644
index 0000000..63d4022
--- /dev/null
+++ b/tests/validation/CL/SpaceToBatchLayer.cpp
@@ -0,0 +1,161 @@
+/*
+ * Copyright (c) 2018 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/core/Types.h"
+#include "arm_compute/runtime/CL/CLTensor.h"
+#include "arm_compute/runtime/CL/CLTensorAllocator.h"
+#include "arm_compute/runtime/CL/functions/CLSpaceToBatchLayer.h"
+#include "tests/CL/CLAccessor.h"
+#include "tests/PaddingCalculator.h"
+#include "tests/datasets/ShapeDatasets.h"
+#include "tests/datasets/SpaceToBatchDataset.h"
+#include "tests/framework/Asserts.h"
+#include "tests/framework/Macros.h"
+#include "tests/framework/datasets/Datasets.h"
+#include "tests/validation/Helpers.h"
+#include "tests/validation/Validation.h"
+#include "tests/validation/fixtures/SpaceToBatchFixture.h"
+
+namespace arm_compute
+{
+namespace test
+{
+namespace validation
+{
+TEST_SUITE(CL)
+TEST_SUITE(SpaceToBatchLayer)
+
+template <typename T>
+using CLSpaceToBatchLayerFixture = SpaceToBatchLayerValidationFixture<CLTensor, CLAccessor, CLSpaceToBatchLayer, T>;
+
+DATA_TEST_CASE(Configuration, framework::DatasetMode::ALL, combine(datasets::SmallSpaceToBatchLayerDataset(),
+                                                                   framework::dataset::make("DataType", { DataType::F16, DataType::F32 })),
+               input_shape, block_shape_shape, paddings_shape, output_shape, dt)
+{
+    // Create tensors
+    CLTensor src         = create_tensor<CLTensor>(input_shape, dt, 1);
+    CLTensor dst         = create_tensor<CLTensor>(output_shape, dt, 1);
+    CLTensor block_shape = create_tensor<CLTensor>(block_shape_shape, DataType::S32, 1);
+    CLTensor paddings    = create_tensor<CLTensor>(paddings_shape, DataType::S32, 1);
+
+    // Create and Configure function
+    CLSpaceToBatchLayer batch_to_space;
+    batch_to_space.configure(&src, &block_shape, &paddings, &dst);
+
+    // Validate valid region
+    const ValidRegion valid_region = shape_to_valid_region(output_shape);
+    validate(dst.info()->valid_region(), valid_region);
+}
+
+// *INDENT-OFF*
+// clang-format off
+DATA_TEST_CASE(Validate, framework::DatasetMode::ALL, zip(zip(zip(zip(
+               framework::dataset::make("InputInfo", { TensorInfo(TensorShape(32U, 13U, 2U, 2U), 1, DataType::F32),
+                                                       TensorInfo(TensorShape(32U, 13U, 2U, 2U), 1, DataType::F32),    // Mismatching data types
+                                                       TensorInfo(TensorShape(32U, 13U, 2U, 2U), 1, DataType::F32),    // Wrong data type block shape
+                                                       TensorInfo(TensorShape(32U, 13U, 2U, 2U, 4U), 1, DataType::F32),    // Wrong tensor shape
+                                                     }),
+               framework::dataset::make("BlockShapeInfo",{ TensorInfo(TensorShape(2U, 2U), 1, DataType::S32),
+                                                       TensorInfo(TensorShape(2U, 2U), 1, DataType::S32),
+                                                       TensorInfo(TensorShape(2U, 2U), 1, DataType::F16),
+                                                       TensorInfo(TensorShape(2U, 2U), 1, DataType::S32),
+                                                     })),
+               framework::dataset::make("PaddingsShapeInfo",{ TensorInfo(TensorShape(2U, 2U), 1, DataType::S32),
+                                                       TensorInfo(TensorShape(2U, 2U), 1, DataType::S32),
+                                                       TensorInfo(TensorShape(2U, 2U), 1, DataType::F16),
+                                                       TensorInfo(TensorShape(2U, 2U), 1, DataType::S32),
+                                                     })),
+               framework::dataset::make("OutputInfo",{ TensorInfo(TensorShape(32U, 13U, 2U, 2U), 1, DataType::F32),
+                                                       TensorInfo(TensorShape(32U, 13U, 2U, 2U), 1, DataType::F16),
+                                                       TensorInfo(TensorShape(32U, 13U, 2U, 2U), 1, DataType::F32),
+                                                       TensorInfo(TensorShape(32U, 13U, 2U, 2U), 1, DataType::F32),
+                                                     })),
+               framework::dataset::make("Expected", { true, false, false, false})),
+               input_info, block_shape_info, paddings_info, output_info, expected)
+{
+    bool has_error = bool(CLSpaceToBatchLayer::validate(&input_info.clone()->set_is_resizable(false), &block_shape_info.clone()->set_is_resizable(false), &paddings_info.clone()->set_is_resizable(false), &output_info.clone()->set_is_resizable(false)));
+    ARM_COMPUTE_EXPECT(has_error == expected, framework::LogLevel::ERRORS);
+}
+DATA_TEST_CASE(ValidateStatic, framework::DatasetMode::ALL, zip(zip(zip(zip(zip(zip(
+               framework::dataset::make("InputInfo", { TensorInfo(TensorShape(32U, 16U, 2U, 1U), 1, DataType::F32),
+                                                       TensorInfo(TensorShape(32U, 16U, 2U, 1U), 1, DataType::F32),    // Mismatching data types
+                                                       TensorInfo(TensorShape(32U, 16U, 2U, 1U), 1, DataType::F32),    // Negative block shapes
+                                                       TensorInfo(TensorShape(32U, 16U, 2U, 1U, 4U), 1, DataType::F32), // Wrong tensor shape
+                                                       TensorInfo(TensorShape(32U, 16U, 2U, 1U, 4U), 1, DataType::F32), // Wrong paddings
+                                                     }),
+               framework::dataset::make("BlockShapeX", { 2, 2, 2, 2, 2 })),
+               framework::dataset::make("BlockShapeY", { 2, 2, -2, 2, 2 })),
+               framework::dataset::make("PadLeft", { Size2D(0, 0), Size2D(0, 0), Size2D(0, 0), Size2D(0, 0), Size2D(3, 11) })),
+               framework::dataset::make("PadRight", { Size2D(0, 0), Size2D(0, 0), Size2D(0, 0), Size2D(0, 0), Size2D(3, 11) })),
+               framework::dataset::make("OutputInfo",{ TensorInfo(TensorShape(16U, 8U, 2U, 4U), 1, DataType::F32),
+                                                       TensorInfo(TensorShape(32U, 8U, 2U, 4U), 1, DataType::F16),
+                                                       TensorInfo(TensorShape(32U, 8U, 2U, 4U), 1, DataType::F32),
+                                                       TensorInfo(TensorShape(32U, 8U, 2U, 4U), 1, DataType::F32),
+                                                       TensorInfo(TensorShape(32U, 8U, 2U, 4U), 1, DataType::F32),
+                                                     })),
+               framework::dataset::make("Expected", { true, false, false, false, false})),
+               input_info, block_shape_x, block_shape_y, padding_left, padding_right, output_info, expected)
+{
+    bool has_error = bool(CLSpaceToBatchLayer::validate(&input_info.clone()->set_is_resizable(false), block_shape_x, block_shape_y, padding_left, padding_right, &output_info.clone()->set_is_resizable(false)));
+    ARM_COMPUTE_EXPECT(has_error == expected, framework::LogLevel::ERRORS);
+}
+// clang-format on
+// *INDENT-ON*
+
+TEST_SUITE(Float)
+TEST_SUITE(FP32)
+FIXTURE_DATA_TEST_CASE(Small, CLSpaceToBatchLayerFixture<float>, framework::DatasetMode::PRECOMMIT, combine(datasets::SmallSpaceToBatchLayerDataset(), framework::dataset::make("DataType",
+                                                                                                            DataType::F32)))
+{
+    // Validate output
+    validate(CLAccessor(_target), _reference);
+}
+FIXTURE_DATA_TEST_CASE(Large, CLSpaceToBatchLayerFixture<float>, framework::DatasetMode::NIGHTLY, combine(datasets::LargeSpaceToBatchLayerDataset(), framework::dataset::make("DataType",
+                                                                                                          DataType::F32)))
+{
+    // Validate output
+    validate(CLAccessor(_target), _reference);
+}
+TEST_SUITE_END() // FP32
+
+TEST_SUITE(FP16)
+FIXTURE_DATA_TEST_CASE(Small, CLSpaceToBatchLayerFixture<half>, framework::DatasetMode::PRECOMMIT, combine(datasets::SmallSpaceToBatchLayerDataset(), framework::dataset::make("DataType",
+                                                                                                           DataType::F16)))
+{
+    // Validate output
+    validate(CLAccessor(_target), _reference);
+}
+FIXTURE_DATA_TEST_CASE(Large, CLSpaceToBatchLayerFixture<half>, framework::DatasetMode::NIGHTLY, combine(datasets::LargeSpaceToBatchLayerDataset(), framework::dataset::make("DataType",
+                                                                                                         DataType::F16)))
+{
+    // Validate output
+    validate(CLAccessor(_target), _reference);
+}
+TEST_SUITE_END() // FP16
+TEST_SUITE_END() // Float
+
+TEST_SUITE_END() // SpaceToBatch
+TEST_SUITE_END() // CL
+} // namespace validation
+} // namespace test
+} // namespace arm_compute
diff --git a/tests/validation/fixtures/SpaceToBatchFixture.h b/tests/validation/fixtures/SpaceToBatchFixture.h
new file mode 100644
index 0000000..26c24c1
--- /dev/null
+++ b/tests/validation/fixtures/SpaceToBatchFixture.h
@@ -0,0 +1,133 @@
+/*
+ * Copyright (c) 2018 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_SPACE_TO_BATCH_LAYER_FIXTURE
+#define ARM_COMPUTE_TEST_SPACE_TO_BATCH_LAYER_FIXTURE
+
+#include "tests/Globals.h"
+#include "tests/framework/Asserts.h"
+#include "tests/framework/Fixture.h"
+#include "tests/validation/reference/SpaceToBatch.h"
+
+namespace arm_compute
+{
+namespace test
+{
+namespace validation
+{
+template <typename TensorType, typename AccessorType, typename FunctionType, typename T>
+class SpaceToBatchLayerValidationFixture : public framework::Fixture
+{
+public:
+    template <typename...>
+    void setup(TensorShape input_shape, TensorShape block_shape_shape, TensorShape paddings_shape, TensorShape output_shape, DataType data_type)
+    {
+        _target    = compute_target(input_shape, block_shape_shape, paddings_shape, output_shape, data_type);
+        _reference = compute_reference(input_shape, block_shape_shape, paddings_shape, output_shape, data_type);
+    }
+
+protected:
+    template <typename U>
+    void fill(U &&tensor, int i)
+    {
+        std::uniform_real_distribution<> distribution(-1.0f, 1.0f);
+        library->fill(tensor, distribution, i);
+    }
+    template <typename U>
+    void fill_pad(U &&tensor, int i)
+    {
+        std::uniform_int_distribution<> distribution(0, 0);
+        library->fill(tensor, distribution, i);
+    }
+    TensorType compute_target(const TensorShape &input_shape, const TensorShape &block_shape_shape, const TensorShape &paddings_shape, const TensorShape &output_shape,
+                              DataType data_type)
+    {
+        // Create tensors
+        TensorType input       = create_tensor<TensorType>(input_shape, data_type);
+        TensorType block_shape = create_tensor<TensorType>(block_shape_shape, DataType::S32);
+        TensorType paddings    = create_tensor<TensorType>(paddings_shape, DataType::S32);
+        TensorType output      = create_tensor<TensorType>(output_shape, data_type);
+
+        // Create and configure function
+        FunctionType space_to_batch;
+        space_to_batch.configure(&input, &block_shape, &paddings, &output);
+
+        ARM_COMPUTE_EXPECT(input.info()->is_resizable(), framework::LogLevel::ERRORS);
+        ARM_COMPUTE_EXPECT(block_shape.info()->is_resizable(), framework::LogLevel::ERRORS);
+        ARM_COMPUTE_EXPECT(paddings.info()->is_resizable(), framework::LogLevel::ERRORS);
+        ARM_COMPUTE_EXPECT(output.info()->is_resizable(), framework::LogLevel::ERRORS);
+
+        // Allocate tensors
+        input.allocator()->allocate();
+        block_shape.allocator()->allocate();
+        paddings.allocator()->allocate();
+        output.allocator()->allocate();
+
+        ARM_COMPUTE_EXPECT(!input.info()->is_resizable(), framework::LogLevel::ERRORS);
+        ARM_COMPUTE_EXPECT(!block_shape.info()->is_resizable(), framework::LogLevel::ERRORS);
+        ARM_COMPUTE_EXPECT(!paddings.info()->is_resizable(), framework::LogLevel::ERRORS);
+        ARM_COMPUTE_EXPECT(!output.info()->is_resizable(), framework::LogLevel::ERRORS);
+
+        // Fill tensors
+        fill(AccessorType(input), 0);
+        fill_pad(AccessorType(paddings), 0);
+        {
+            auto block_shape_data = AccessorType(block_shape);
+            for(unsigned int i = 0; i < block_shape_shape.x(); ++i)
+            {
+                static_cast<int32_t *>(block_shape_data.data())[i] = input_shape[i] / output_shape[i];
+            }
+        }
+        // Compute function
+        space_to_batch.run();
+
+        return output;
+    }
+
+    SimpleTensor<T> compute_reference(const TensorShape &input_shape, const TensorShape &block_shape_shape, const TensorShape &paddings_shape,
+                                      const TensorShape &output_shape, DataType data_type)
+    {
+        // Create reference
+        SimpleTensor<T>       input{ input_shape, data_type };
+        SimpleTensor<int32_t> block_shape{ block_shape_shape, DataType::S32 };
+        SimpleTensor<int32_t> paddings{ paddings_shape, DataType::S32 };
+
+        // Fill reference
+        fill(input, 0);
+        fill_pad(paddings, 0);
+        for(unsigned int i = 0; i < block_shape_shape.x(); ++i)
+        {
+            block_shape[i] = input_shape[i] / output_shape[i];
+        }
+
+        // Compute reference
+        return reference::space_to_batch(input, block_shape, paddings, output_shape);
+    }
+
+    TensorType      _target{};
+    SimpleTensor<T> _reference{};
+};
+} // namespace validation
+} // namespace test
+} // namespace arm_compute
+#endif /* ARM_COMPUTE_TEST_SPACE_TO_BATCH_LAYER_FIXTURE */
diff --git a/tests/validation/reference/SpaceToBatch.cpp b/tests/validation/reference/SpaceToBatch.cpp
new file mode 100644
index 0000000..979ab94
--- /dev/null
+++ b/tests/validation/reference/SpaceToBatch.cpp
@@ -0,0 +1,82 @@
+/*
+ * Copyright (c) 2018 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 "SpaceToBatch.h"
+
+#include "tests/validation/Helpers.h"
+
+namespace arm_compute
+{
+namespace test
+{
+namespace validation
+{
+namespace reference
+{
+// Space to Batch
+template <typename T>
+SimpleTensor<T> space_to_batch(const SimpleTensor<T> &src, const SimpleTensor<int32_t> &block_shape, const SimpleTensor<int32_t> &paddings, const TensorShape &dst_shape)
+{
+    SimpleTensor<T> result(dst_shape, src.data_type());
+
+    auto width_out  = static_cast<int>(dst_shape[0]);
+    auto height_out = static_cast<int>(dst_shape[1]);
+    auto z_out      = static_cast<int>(dst_shape[2]);
+
+    int out_pos = 0;
+    for(int batch = 0; batch < static_cast<int>(dst_shape[3]); ++batch)
+    {
+        for(int z = 0; z < z_out; ++z)
+        {
+            for(int y = 0; y < height_out; ++y)
+            {
+                for(int x = 0; x < width_out; ++x)
+                {
+                    if(x < paddings[0] || x > width_out - paddings[1] - 1
+                       || y < paddings[2] || y > height_out - paddings[3] - 1)
+                    {
+                        result[out_pos] = 0;
+                    }
+                    else
+                    {
+                        const int r      = dst_shape[3] / (block_shape[0] * block_shape[1]);
+                        const int in_x   = (block_shape[0] * (x - paddings[0]) + (batch / r) % block_shape[0]);
+                        const int in_y   = (block_shape[1] * (y - paddings[2]) + (batch / r) / block_shape[0]);
+                        int       in_pos = in_x + src.shape()[0] * in_y + z * src.shape()[0] * src.shape()[1] + (batch % r) * src.shape()[0] * src.shape()[1] * src.shape()[2];
+                        result[out_pos]  = src[in_pos];
+                    }
+                    ++out_pos;
+                }
+            }
+        }
+    }
+
+    return result;
+}
+
+template SimpleTensor<float> space_to_batch(const SimpleTensor<float> &src, const SimpleTensor<int32_t> &block_shape, const SimpleTensor<int32_t> &paddings, const TensorShape &dst_shape);
+template SimpleTensor<half> space_to_batch(const SimpleTensor<half> &src, const SimpleTensor<int32_t> &block_shape, const SimpleTensor<int32_t> &paddings, const TensorShape &dst_shape);
+} // namespace reference
+} // namespace validation
+} // namespace test
+} // namespace arm_compute
diff --git a/tests/validation/reference/SpaceToBatch.h b/tests/validation/reference/SpaceToBatch.h
new file mode 100644
index 0000000..ba35211
--- /dev/null
+++ b/tests/validation/reference/SpaceToBatch.h
@@ -0,0 +1,44 @@
+/*
+ * Copyright (c) 2018 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_SPACE_TO_BATCH_LAYER_H__
+#define __ARM_COMPUTE_TEST_SPACE_TO_BATCH_LAYER_H__
+
+#include "tests/SimpleTensor.h"
+#include "tests/validation/Helpers.h"
+
+namespace arm_compute
+{
+namespace test
+{
+namespace validation
+{
+namespace reference
+{
+template <typename T>
+SimpleTensor<T> space_to_batch(const SimpleTensor<T> &src, const SimpleTensor<int32_t> &block_shape, const SimpleTensor<int32_t> &paddings, const TensorShape &dst_shape);
+} // namespace reference
+} // namespace validation
+} // namespace test
+} // namespace arm_compute
+#endif /* __ARM_COMPUTE_TEST_SPACE_TO_BATCH_LAYER_H__ */