COMPMID-1946: Implement NEBatchToSpace

Change-Id: I119645eb3ea437c7dfe59545da58b328a7184f3f
Signed-off-by: giuros01 <giuseppe.rossini@arm.com>
Reviewed-on: https://review.mlplatform.org/c/734
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com>

7 files changed, 584 insertions, 0 deletions

diff --git a/arm_compute/core/NEON/NEKernels.h b/arm_compute/core/NEON/NEKernels.h
index 6509e39d7d..8b37b2f603 100644
--- a/arm_compute/core/NEON/NEKernels.h
+++ b/arm_compute/core/NEON/NEKernels.h
@@ -31,6 +31,7 @@
 #include "arm_compute/core/NEON/kernels/NEArithmeticAdditionKernel.h"
 #include "arm_compute/core/NEON/kernels/NEArithmeticSubtractionKernel.h"
 #include "arm_compute/core/NEON/kernels/NEBatchNormalizationLayerKernel.h"
+#include "arm_compute/core/NEON/kernels/NEBatchToSpaceLayerKernel.h"
 #include "arm_compute/core/NEON/kernels/NEBitwiseAndKernel.h"
 #include "arm_compute/core/NEON/kernels/NEBitwiseNotKernel.h"
 #include "arm_compute/core/NEON/kernels/NEBitwiseOrKernel.h"
diff --git a/arm_compute/core/NEON/kernels/NEBatchToSpaceLayerKernel.h b/arm_compute/core/NEON/kernels/NEBatchToSpaceLayerKernel.h
new file mode 100644
index 0000000000..7a76e88576
--- /dev/null
+++ b/arm_compute/core/NEON/kernels/NEBatchToSpaceLayerKernel.h
@@ -0,0 +1,100 @@
+/*
+ * Copyright (c) 2019 ARM Limited.
+ *
+ * SPDX-License-Identifier: MIT
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to
+ * deal in the Software without restriction, including without limitation the
+ * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or
+ * sell copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in all
+ * copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+#ifndef __ARM_COMPUTE_NEBATCHTOSPACELAYERKERNEL_H__
+#define __ARM_COMPUTE_NEBATCHTOSPACELAYERKERNEL_H__
+
+#include "arm_compute/core/NEON/INEKernel.h"
+
+namespace arm_compute
+{
+class ITensor;
+
+/** Interface for the batch to space kernel */
+class NEBatchToSpaceLayerKernel : public INEKernel
+{
+public:
+    const char *name() const override
+    {
+        return "NEBatchToSpaceLayerKernel";
+    }
+    /** Default constructor */
+    NEBatchToSpaceLayerKernel();
+    /** Prevent instances of this class from being copied (As this class contains pointers) */
+    NEBatchToSpaceLayerKernel(const NEBatchToSpaceLayerKernel &) = delete;
+    /** Prevent instances of this class from being copied (As this class contains pointers) */
+    NEBatchToSpaceLayerKernel &operator=(const NEBatchToSpaceLayerKernel &) = delete;
+    /** Allow instances of this class to be moved */
+    NEBatchToSpaceLayerKernel(NEBatchToSpaceLayerKernel &&) = default;
+    /** Allow instances of this class to be moved */
+    NEBatchToSpaceLayerKernel &operator=(NEBatchToSpaceLayerKernel &&) = default;
+    /** Default destructor */
+    ~NEBatchToSpaceLayerKernel() = default;
+    /** Initialise the kernel's inputs and output.
+     *
+     * @param[in]  input       Tensor input. Supported tensor rank: 4. Data types supported: U8/S8/QASYMM8/U16/S16/F16/U32/S32/F32.
+     * @param[in]  block_shape 1-D tensor with shape [M]. Data types supported: S32
+     * @param[out] output      Tensor output. Data types supported: same as @p input
+     */
+    void configure(const ITensor *input, const ITensor *block_shape, ITensor *output);
+    /** Initialise the kernel's inputs and output (Static block shape).
+     *
+     * @param[in]  input         Tensor input. Supported tensor rank: 4. Data types supported: U8/S8/QASYMM8/U16/S16/F16/U32/S32/F32.
+     * @param[in]  block_shape_x Block shape x value.
+     * @param[in]  block_shape_y Block shape y value.
+     * @param[out] output        Tensor output. Data types supported: same as @p input
+     */
+    void configure(const ITensor *input, const int32_t block_shape_x, const int32_t block_shape_y, ITensor *output);
+    /** Static function to check if given info will lead to a valid configuration of @ref NEBatchToSpaceLayerKernel
+     *
+     * @param[in] input       Tensor input. Supported tensor rank: 4. Data types supported: U8/S8/QASYMM8/U16/S16/F16/U32/S32/F32.
+     * @param[in] block_shape 1-D tensor with shape [M]. Data types supported: S32
+     * @param[in] output      Tensor output. Data types supported: same as @p input
+     *
+     * @return a status
+     */
+    static Status validate(const ITensorInfo *input, const ITensorInfo *block_shape, const ITensorInfo *output);
+    /** Static function to check if given info will lead to a valid configuration of @ref NEBatchToSpaceLayerKernel (Static block shape).
+     *
+     * @param[in] input         Tensor input. Supported tensor rank: 4. Data types supported: U8/S8/QASYMM8/U16/S16/F16/U32/S32/F32.
+     * @param[in] block_shape_x Block shape x value.
+     * @param[in] block_shape_y Block shape y value.
+     * @param[in] output        Tensor output. Data types supported: same as @p input
+     *
+     * @return a status
+     */
+    static Status validate(const ITensorInfo *input, const int32_t block_shape_x, const int32_t block_shape_y, const ITensorInfo *output);
+
+    // Inherited methods overridden:
+    void run(const Window &window, const ThreadInfo &info) override;
+
+private:
+    const ITensor *_input;       /**< Source tensor */
+    const ITensor *_block_shape; /**< Block shape tensor */
+    ITensor       *_output;      /**< Destination tensor */
+
+    int32_t _block_shape_x;
+    int32_t _block_shape_y;
+};
+} // namespace arm_compute
+#endif /* __ARM_COMPUTE_NEBATCHTOSPACELAYERKERNEL_H__ */
diff --git a/arm_compute/runtime/NEON/NEFunctions.h b/arm_compute/runtime/NEON/NEFunctions.h
index a07e432b90..15ce4e3d66 100644
--- a/arm_compute/runtime/NEON/NEFunctions.h
+++ b/arm_compute/runtime/NEON/NEFunctions.h
@@ -32,6 +32,7 @@
 #include "arm_compute/runtime/NEON/functions/NEArithmeticAddition.h"
 #include "arm_compute/runtime/NEON/functions/NEArithmeticSubtraction.h"
 #include "arm_compute/runtime/NEON/functions/NEBatchNormalizationLayer.h"
+#include "arm_compute/runtime/NEON/functions/NEBatchToSpaceLayer.h"
 #include "arm_compute/runtime/NEON/functions/NEBitwiseAnd.h"
 #include "arm_compute/runtime/NEON/functions/NEBitwiseNot.h"
 #include "arm_compute/runtime/NEON/functions/NEBitwiseOr.h"
diff --git a/arm_compute/runtime/NEON/functions/NEBatchToSpaceLayer.h b/arm_compute/runtime/NEON/functions/NEBatchToSpaceLayer.h
new file mode 100644
index 0000000000..3ab3fa112e
--- /dev/null
+++ b/arm_compute/runtime/NEON/functions/NEBatchToSpaceLayer.h
@@ -0,0 +1,77 @@
+/*
+ * Copyright (c) 2019 ARM Limited.
+ *
+ * SPDX-License-Identifier: MIT
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to
+ * deal in the Software without restriction, including without limitation the
+ * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or
+ * sell copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in all
+ * copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+#ifndef __ARM_COMPUTE_NEBATCHTOSPACELAYER_H__
+#define __ARM_COMPUTE_NEBATCHTOSPACELAYER_H__
+
+#include "arm_compute/runtime/IFunction.h"
+
+#include "arm_compute/core/NEON/kernels/NEBatchToSpaceLayerKernel.h"
+#include "arm_compute/core/Types.h"
+#include "arm_compute/runtime/NEON/INESimpleFunctionNoBorder.h"
+
+namespace arm_compute
+{
+class ITensor;
+
+/** Basic function to run @ref NEBatchToSpaceLayerKernel. */
+class NEBatchToSpaceLayer : public INESimpleFunctionNoBorder
+{
+public:
+    /** Set the input and output tensors.
+     *
+     * @param[in]  input       Tensor input. Supported tensor rank: 4. Data types supported: U8/S8/QASYMM8/U16/S16/F16/U32/S32/F32.
+     * @param[in]  block_shape 1-D tensor with shape [M]. Data types supported: S32
+     * @param[out] output      Tensor output. Data types supported: same as @p input
+     */
+    void configure(const ITensor *input, const ITensor *block_shape, ITensor *output);
+    /** Set the input and output tensors. (Static block shape).
+     *
+     * @param[in]  input         Tensor input. Supported tensor rank: 4. Data types supported: U8/S8/QASYMM8/U16/S16/F16/U32/S32/F32.
+     * @param[in]  block_shape_x Block shape x value.
+     * @param[in]  block_shape_y Block shape y value.
+     * @param[out] output        Tensor output. Data types supported: same as @p input
+     */
+    void configure(const ITensor *input, int32_t block_shape_x, int32_t block_shape_y, ITensor *output);
+    /** Static function to check if given info will lead to a valid configuration of @ref CLBatchToSpaceLayer
+     *
+     * @param[in]  input       Tensor input info. Supported tensor rank: 4. Data types supported: U8/S8/QASYMM8/U16/S16/F16/U32/S32/F32.
+     * @param[in]  block_shape block shape tensor info with shape [M]. Data types supported: S32
+     * @param[out] output      Tensor output info. Data types supported: same as @p input
+     *
+     * @return a status
+     */
+    static Status validate(const ITensorInfo *input, const ITensorInfo *block_shape, const ITensorInfo *output);
+    /** Static function to check if given info will lead to a valid configuration of @ref CLBatchToSpaceLayer (Static block shape).
+     *
+     * @param[in]  input         Tensor input info. Supported tensor rank: 4. Data types supported: U8/S8/QASYMM8/U16/S16/F16/U32/S32/F32.
+     * @param[in]  block_shape_x Block shape x value.
+     * @param[in]  block_shape_y Block shape y value.
+     * @param[out] output        Tensor output info. Data types supported: same as @p input
+     *
+     * @return a status
+     */
+    static Status validate(const ITensorInfo *input, int32_t block_shape_x, int32_t block_shape_y, const ITensorInfo *output);
+};
+} // namespace arm_compute
+#endif /* __ARM_COMPUTE_NEBATCHTOSPACELAYER_H__ */
diff --git a/src/core/NEON/kernels/NEBatchToSpaceLayerKernel.cpp b/src/core/NEON/kernels/NEBatchToSpaceLayerKernel.cpp
new file mode 100644
index 0000000000..8c24d9fad8
--- /dev/null
+++ b/src/core/NEON/kernels/NEBatchToSpaceLayerKernel.cpp
@@ -0,0 +1,212 @@
+/*
+ * Copyright (c) 2019 ARM Limited.
+ *
+ * SPDX-License-Identifier: MIT
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to
+ * deal in the Software without restriction, including without limitation the
+ * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or
+ * sell copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in all
+ * copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+#include "arm_compute/core/NEON/kernels/NEBatchToSpaceLayerKernel.h"
+
+#include "arm_compute/core/Helpers.h"
+#include "arm_compute/core/ITensor.h"
+#include "arm_compute/core/NEON/wrapper/wrapper.h"
+#include "arm_compute/core/Types.h"
+#include "arm_compute/core/Validate.h"
+#include "arm_compute/core/utils/misc/ShapeCalculator.h"
+#include <arm_neon.h>
+#include <cstdint>
+
+using namespace arm_compute::misc::shape_calculator;
+
+namespace arm_compute
+{
+namespace
+{
+Status validate_arguments(const ITensorInfo *input, const ITensorInfo *block_info, const ITensorInfo *output)
+{
+    ARM_COMPUTE_RETURN_ERROR_ON_NULLPTR(input, block_info, output);
+    ARM_COMPUTE_RETURN_ERROR_ON_DATA_TYPE_CHANNEL_NOT_IN(block_info, 1, DataType::S32);
+    ARM_COMPUTE_RETURN_ERROR_ON(input->num_dimensions() > 4);
+
+    // Validate output if initialized
+    if(output->total_size() != 0)
+    {
+        ARM_COMPUTE_RETURN_ERROR_ON(output->num_dimensions() > 4);
+        ARM_COMPUTE_RETURN_ERROR_ON_MISMATCHING_DATA_TYPES(input, output);
+    }
+
+    return Status{};
+}
+Status validate_arguments_static(const ITensorInfo *input, const int block_shape_x, const int block_shape_y, const ITensorInfo *output)
+{
+    ARM_COMPUTE_RETURN_ERROR_ON_NULLPTR(input, output);
+    ARM_COMPUTE_RETURN_ERROR_ON(input->num_dimensions() > 4);
+    ARM_COMPUTE_RETURN_ERROR_ON(block_shape_x <= 0);
+    ARM_COMPUTE_RETURN_ERROR_ON(block_shape_y <= 0);
+
+    const DataLayout data_layout = input->data_layout();
+    const int        idx_batch   = get_data_layout_dimension_index(data_layout, DataLayoutDimension::BATCHES);
+    ARM_COMPUTE_RETURN_ERROR_ON(input->tensor_shape()[idx_batch] % (block_shape_x * block_shape_y) != 0);
+
+    // Validate output if initialized
+    if(output->total_size() != 0)
+    {
+        const int idx_width   = get_data_layout_dimension_index(data_layout, DataLayoutDimension::WIDTH);
+        const int idx_height  = get_data_layout_dimension_index(data_layout, DataLayoutDimension::HEIGHT);
+        const int idx_channel = get_data_layout_dimension_index(data_layout, DataLayoutDimension::CHANNEL);
+        ARM_COMPUTE_RETURN_ERROR_ON(output->tensor_shape()[idx_width] != (block_shape_x * input->tensor_shape()[idx_width]));
+        ARM_COMPUTE_RETURN_ERROR_ON(output->tensor_shape()[idx_height] != (block_shape_x * input->tensor_shape()[idx_height]));
+        ARM_COMPUTE_RETURN_ERROR_ON(output->tensor_shape()[idx_channel] != input->tensor_shape()[idx_channel]);
+        ARM_COMPUTE_RETURN_ERROR_ON(output->num_dimensions() > 4);
+        ARM_COMPUTE_RETURN_ERROR_ON_MISMATCHING_DATA_TYPES(input, output);
+    }
+
+    return Status{};
+}
+} // namespace
+
+NEBatchToSpaceLayerKernel::NEBatchToSpaceLayerKernel()
+    : _input(nullptr), _block_shape(nullptr), _output(nullptr), _block_shape_x(), _block_shape_y()
+{
+}
+
+void NEBatchToSpaceLayerKernel::configure(const ITensor *input, const ITensor *block_shape, ITensor *output)
+{
+    ARM_COMPUTE_ERROR_ON_NULLPTR(input, output);
+    ARM_COMPUTE_ERROR_THROW_ON(validate_arguments(input->info(), block_shape->info(), output->info()));
+
+    _input       = input;
+    _block_shape = block_shape;
+    _output      = output;
+
+    // Configure kernel window
+    Window win = calculate_max_window(*input->info(), Steps());
+    ICPPKernel::configure(win);
+}
+
+void NEBatchToSpaceLayerKernel::configure(const ITensor *input, const int32_t block_shape_x, const int32_t block_shape_y, ITensor *output)
+{
+    ARM_COMPUTE_ERROR_ON_NULLPTR(input, output);
+    TensorShape output_shape = compute_batch_to_space_shape(input->info(), block_shape_x, block_shape_y);
+    // Output auto inizialitation if not yet initialized
+    auto_init_if_empty(*output->info(), input->info()->clone()->set_tensor_shape(output_shape));
+
+    // Perform validation step
+    ARM_COMPUTE_ERROR_THROW_ON(validate_arguments_static(input->info(), block_shape_x, block_shape_y, output->info()));
+
+    _input         = input;
+    _output        = output;
+    _block_shape_x = block_shape_x;
+    _block_shape_y = block_shape_y;
+
+    // Configure kernel window
+    Window win = calculate_max_window(*input->info(), Steps());
+    ICPPKernel::configure(win);
+}
+
+Status NEBatchToSpaceLayerKernel::validate(const ITensorInfo *input, const ITensorInfo *block_shape, const ITensorInfo *output)
+{
+    ARM_COMPUTE_RETURN_ERROR_ON_NULLPTR(input, block_shape, output);
+    ARM_COMPUTE_RETURN_ON_ERROR(validate_arguments(input, block_shape, output));
+    return Status{};
+}
+
+Status NEBatchToSpaceLayerKernel::validate(const ITensorInfo *input, const int32_t block_shape_x, const int32_t block_shape_y, const ITensorInfo *output)
+{
+    ARM_COMPUTE_RETURN_ERROR_ON_NULLPTR(input, output);
+    ARM_COMPUTE_RETURN_ON_ERROR(validate_arguments_static(input, block_shape_x, block_shape_y, output));
+    return Status{};
+}
+
+void NEBatchToSpaceLayerKernel::run(const Window &window, const ThreadInfo &info)
+{
+    ARM_COMPUTE_UNUSED(info);
+    ARM_COMPUTE_ERROR_ON_UNCONFIGURED_KERNEL(this);
+    ARM_COMPUTE_ERROR_ON_INVALID_SUBWINDOW(ICPPKernel::window(), window);
+
+    if(_block_shape != nullptr)
+    {
+        // Retrieve the block shapes dynamically
+        _block_shape_x = *(reinterpret_cast<const int *>(_block_shape->ptr_to_element(0)));
+        _block_shape_y = *(reinterpret_cast<const int *>(_block_shape->ptr_to_element(1)));
+    }
+
+    const int batch_size   = _input->info()->dimension(3);
+    const int r            = (batch_size / (_block_shape_x * _block_shape_y));
+    const int element_size = _input->info()->element_size();
+
+    Window slice_in  = window.first_slice_window_3D();
+    Window slice_out = window.first_slice_window_4D();
+
+    // The slice_out slice does not move
+    slice_out.set(Window::DimX, Window::Dimension(0, 0, 0));
+    slice_out.set(Window::DimY, Window::Dimension(0, 0, 0));
+    slice_out.set(Window::DimZ, Window::Dimension(0, 0, 0));
+    slice_out.set(3, Window::Dimension(0, 0, 0));
+
+    int batch_id = 0;
+    // Main loop for NCHW and NHWC
+    if(_input->info()->data_layout() == DataLayout::NCHW)
+    {
+        do
+        {
+            Iterator in(_input, slice_in);
+            execute_window_loop(slice_in, [&](const Coordinates & id)
+            {
+
+                const int x = id.x();
+                const int y = id.y();
+                const int z = id.z();
+
+                const int   w     = batch_id % r;
+                const int   out_x = x * _block_shape_x + (batch_id / r) % _block_shape_x;
+                const int   out_y = y * _block_shape_y + (batch_id / r) / _block_shape_x;
+                Coordinates output_coords{ out_x, out_y, z, w };
+                memcpy(_output->ptr_to_element(output_coords), in.ptr(), element_size);
+            },
+            in);
+            ++batch_id;
+        }
+        while(window.slide_window_slice_3D(slice_in));
+    }
+    else
+    {
+        do
+        {
+            Iterator in(_input, slice_in);
+            execute_window_loop(slice_in, [&](const Coordinates & id)
+            {
+
+                const int z = id.x();
+                const int x = id.y();
+                const int y = id.z();
+
+                const int   w     = batch_id % r;
+                const int   out_x = x * _block_shape_x + (batch_id / r) % _block_shape_x;
+                const int   out_y = y * _block_shape_y + (batch_id / r) / _block_shape_x;
+                Coordinates output_coords{ z, out_x, out_y, w };
+                memcpy(_output->ptr_to_element(output_coords), in.ptr(), element_size);
+            },
+            in);
+            ++batch_id;
+        }
+        while(window.slide_window_slice_3D(slice_in));
+    }
+}
+} // namespace arm_compute
diff --git a/src/runtime/NEON/functions/NEBatchToSpaceLayer.cpp b/src/runtime/NEON/functions/NEBatchToSpaceLayer.cpp
new file mode 100644
index 0000000000..a4db1fdda3
--- /dev/null
+++ b/src/runtime/NEON/functions/NEBatchToSpaceLayer.cpp
@@ -0,0 +1,57 @@
+/*
+ * Copyright (c) 2019 ARM Limited.
+ *
+ * SPDX-License-Identifier: MIT
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to
+ * deal in the Software without restriction, including without limitation the
+ * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or
+ * sell copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in all
+ * copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+
+#include "arm_compute/runtime/NEON/functions/NEBatchToSpaceLayer.h"
+
+#include "arm_compute/core/Error.h"
+#include "arm_compute/core/TensorInfo.h"
+#include "arm_compute/core/Types.h"
+#include "arm_compute/core/Validate.h"
+
+namespace arm_compute
+{
+void NEBatchToSpaceLayer::configure(const ITensor *input, const ITensor *block_shape, ITensor *output)
+{
+    auto k = arm_compute::support::cpp14::make_unique<NEBatchToSpaceLayerKernel>();
+    k->configure(input, block_shape, output);
+    _kernel = std::move(k);
+}
+
+void NEBatchToSpaceLayer::configure(const ITensor *input, int32_t block_shape_x, int32_t block_shape_y, ITensor *output)
+{
+    auto k = arm_compute::support::cpp14::make_unique<NEBatchToSpaceLayerKernel>();
+    k->configure(input, block_shape_x, block_shape_y, output);
+    _kernel = std::move(k);
+}
+
+Status NEBatchToSpaceLayer::validate(const ITensorInfo *input, const ITensorInfo *block_shape, const ITensorInfo *output)
+{
+    return NEBatchToSpaceLayerKernel::validate(input, block_shape, output);
+}
+
+Status NEBatchToSpaceLayer::validate(const ITensorInfo *input, int32_t block_shape_x, int32_t block_shape_y, const ITensorInfo *output)
+{
+    return NEBatchToSpaceLayerKernel::validate(input, block_shape_x, block_shape_y, output);
+}
+} // namespace arm_compute
diff --git a/tests/validation/NEON/BatchToSpaceLayer.cpp b/tests/validation/NEON/BatchToSpaceLayer.cpp
new file mode 100644
index 0000000000..96ea5b8241
--- /dev/null
+++ b/tests/validation/NEON/BatchToSpaceLayer.cpp
@@ -0,0 +1,136 @@
+/*
+ * Copyright (c) 2019 ARM Limited.
+ *
+ * SPDX-License-Identifier: MIT
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to
+ * deal in the Software without restriction, including without limitation the
+ * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or
+ * sell copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in all
+ * copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+#include "arm_compute/core/Types.h"
+#include "arm_compute/runtime/NEON/functions/NEBatchToSpaceLayer.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/BatchToSpaceDataset.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/BatchToSpaceLayerFixture.h"
+
+namespace arm_compute
+{
+namespace test
+{
+namespace validation
+{
+TEST_SUITE(NEON)
+TEST_SUITE(BatchToSpaceLayer)
+
+template <typename T>
+using NEBatchToSpaceLayerFixture = BatchToSpaceLayerValidationFixture<Tensor, Accessor, NEBatchToSpaceLayer, T>;
+
+// *INDENT-OFF*
+// clang-format off
+DATA_TEST_CASE(Validate, framework::DatasetMode::ALL, 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("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, output_info, expected)
+{
+    bool has_error = bool(NEBatchToSpaceLayer::validate(&input_info.clone()->set_is_resizable(false), &block_shape_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(
+               framework::dataset::make("InputInfo", { TensorInfo(TensorShape(16U, 8U, 2U, 4U), 1, DataType::F32),
+                                                       TensorInfo(TensorShape(16U, 8U, 2U, 4U), 1, DataType::F32),    // Mismatching data types
+                                                       TensorInfo(TensorShape(16U, 8U, 2U, 4U), 1, DataType::F32),    // Negative block shapes
+                                                       TensorInfo(TensorShape(32U, 16U, 2U, 4U, 4U), 1, DataType::F32), // Wrong tensor shape
+                                                     }),
+               framework::dataset::make("BlockShapeX", { 2, 2, 2, 2 })),
+               framework::dataset::make("BlockShapeY", { 2, 2, -2, 2 })),
+               framework::dataset::make("OutputInfo",{ TensorInfo(TensorShape(32U, 16U, 2U, 1U), 1, DataType::F32),
+                                                       TensorInfo(TensorShape(32U, 16U, 2U, 1U), 1, DataType::F16),
+                                                       TensorInfo(TensorShape(32U, 16U, 2U, 1U), 1, DataType::F32),
+                                                       TensorInfo(TensorShape(32U, 8U, 2U, 1U), 1, DataType::F32),
+                                                     })),
+               framework::dataset::make("Expected", { true, false, false, false})),
+               input_info, block_shape_x, block_shape_y, output_info, expected)
+{
+    bool has_error = bool(NEBatchToSpaceLayer::validate(&input_info.clone()->set_is_resizable(false), block_shape_x, block_shape_y, &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(RunSmall, NEBatchToSpaceLayerFixture<float>, framework::DatasetMode::PRECOMMIT, combine(combine(datasets::SmallBatchToSpaceLayerDataset(), framework::dataset::make("DataType",
+                                                                                                                       DataType::F32)),
+                                                                                                               framework::dataset::make("DataLayout", { DataLayout::NCHW, DataLayout::NHWC })))
+{
+    // Validate output
+    validate(Accessor(_target), _reference);
+}
+FIXTURE_DATA_TEST_CASE(RunLarge, NEBatchToSpaceLayerFixture<float>, framework::DatasetMode::NIGHTLY, combine(combine(datasets::LargeBatchToSpaceLayerDataset(), framework::dataset::make("DataType",
+                                                                                                                     DataType::F32)),
+                                                                                                             framework::dataset::make("DataLayout", { DataLayout::NCHW, DataLayout::NHWC })))
+{
+    // Validate output
+    validate(Accessor(_target), _reference);
+}
+TEST_SUITE_END()
+
+TEST_SUITE(FP16)
+FIXTURE_DATA_TEST_CASE(RunSmall, NEBatchToSpaceLayerFixture<half>, framework::DatasetMode::PRECOMMIT, combine(combine(datasets::SmallBatchToSpaceLayerDataset(), framework::dataset::make("DataType",
+                                                                                                                      DataType::F16)),
+                                                                                                              framework::dataset::make("DataLayout", { DataLayout::NCHW, DataLayout::NHWC })))
+{
+    // Validate output
+    validate(Accessor(_target), _reference);
+}
+FIXTURE_DATA_TEST_CASE(RunLarge, NEBatchToSpaceLayerFixture<half>, framework::DatasetMode::NIGHTLY, combine(combine(datasets::LargeBatchToSpaceLayerDataset(), framework::dataset::make("DataType",
+                                                                                                                    DataType::F16)),
+                                                                                                            framework::dataset::make("DataLayout", { DataLayout::NCHW, DataLayout::NHWC })))
+{
+    // Validate output
+    validate(Accessor(_target), _reference);
+}
+TEST_SUITE_END()
+TEST_SUITE_END()
+
+TEST_SUITE_END() // BatchToSpace
+TEST_SUITE_END() // NEON
+} // namespace validation
+} // namespace test
+} // namespace arm_compute