diff --git a/src/core/CL/kernels/CLDepthwiseConvolutionLayerNativeKernel.cpp b/src/core/CL/kernels/CLDepthwiseConvolutionLayerNativeKernel.cpp
index 8b624bb..2115fc6 100644
--- a/src/core/CL/kernels/CLDepthwiseConvolutionLayerNativeKernel.cpp
+++ b/src/core/CL/kernels/CLDepthwiseConvolutionLayerNativeKernel.cpp
@@ -35,6 +35,7 @@
 #include "arm_compute/core/Types.h"
 #include "arm_compute/core/Utils.h"
 #include "arm_compute/core/utils/misc/ShapeCalculator.h"
+#include "arm_compute/core/utils/quantization/AsymmHelpers.h"
 
 namespace arm_compute
 {
@@ -46,7 +47,7 @@
     ARM_COMPUTE_UNUSED(dwc_info);
     ARM_COMPUTE_RETURN_ERROR_ON_F16_UNSUPPORTED(input);
     ARM_COMPUTE_RETURN_ERROR_ON_DATA_LAYOUT_NOT_IN(input, DataLayout::NHWC);
-    ARM_COMPUTE_RETURN_ERROR_ON_DATA_TYPE_CHANNEL_NOT_IN(input, 1, DataType::F16, DataType::F32);
+    ARM_COMPUTE_RETURN_ERROR_ON_DATA_TYPE_CHANNEL_NOT_IN(input, 1, DataType::QASYMM8, DataType::F16, DataType::F32);
     ARM_COMPUTE_RETURN_ERROR_ON_MISMATCHING_DATA_TYPES(input, weights);
     ARM_COMPUTE_RETURN_ERROR_ON(depth_multiplier > 1 && dwc_weights_info.n0 != 1);
     ARM_COMPUTE_RETURN_ERROR_ON(conv_info.stride().first < 1);
@@ -59,8 +60,16 @@
     if(biases != nullptr)
     {
         ARM_COMPUTE_RETURN_ERROR_ON(biases->dimension(0) != weights->dimension(0));
-        ARM_COMPUTE_RETURN_ERROR_ON_MISMATCHING_DATA_TYPES(weights, biases);
         ARM_COMPUTE_RETURN_ERROR_ON(biases->num_dimensions() > 1);
+
+        if(is_data_type_quantized(input->data_type()))
+        {
+            ARM_COMPUTE_RETURN_ERROR_ON_DATA_TYPE_CHANNEL_NOT_IN(biases, 1, DataType::S32);
+        }
+        else
+        {
+            ARM_COMPUTE_RETURN_ERROR_ON_MISMATCHING_DATA_TYPES(weights, biases);
+        }
     }
 
     if(output->total_size() != 0)
@@ -137,6 +146,7 @@
     const size_t idx_h          = get_data_layout_dimension_index(input->info()->data_layout(), DataLayoutDimension::HEIGHT);
     const size_t weights_width  = weights->info()->dimension(idx_w);
     const size_t weights_height = weights->info()->dimension(idx_h);
+    const bool   is_quantized   = is_data_type_quantized(input->info()->data_type());
 
     CLBuildOptions build_opts;
     build_opts.add_option_if(_biases != nullptr, "-DHAS_BIAS");
@@ -155,10 +165,46 @@
     build_opts.add_option("-DCONV_STRIDE_Y=" + support::cpp11::to_string(conv_info.stride().second));
     build_opts.add_option("-DDILATION_X=" + support::cpp11::to_string(dilation.x()));
     build_opts.add_option("-DDILATION_Y=" + support::cpp11::to_string(dilation.y()));
-    build_opts.add_option_if(dwc_info.activation_info.enabled(), "-DA_VAL=" + float_to_string_with_full_precision(dwc_info.activation_info.a()));
-    build_opts.add_option_if(dwc_info.activation_info.enabled(), "-DB_VAL=" + float_to_string_with_full_precision(dwc_info.activation_info.b()));
 
-    std::string kernel_name("dwc_MxN_native_fp_nhwc");
+    std::string kernel_name = (is_quantized) ? "dwc_MxN_native_quantized8_nhwc" : "dwc_MxN_native_fp_nhwc";
+
+    if(is_quantized)
+    {
+        const UniformQuantizationInfo iq_info = _input->info()->quantization_info().uniform();
+        const UniformQuantizationInfo wq_info = _weights->info()->quantization_info().uniform();
+        const UniformQuantizationInfo oq_info = _output->info()->quantization_info().uniform();
+
+        float multiplier        = iq_info.scale * wq_info.scale / oq_info.scale;
+        int   output_multiplier = 0;
+        int   output_shift      = 0;
+        quantization::calculate_quantized_multiplier_less_than_one(multiplier, &output_multiplier, &output_shift);
+
+        build_opts.add_option("-DINPUT_OFFSET=" + support::cpp11::to_string(-iq_info.offset));
+        build_opts.add_option("-DWEIGHTS_OFFSET=" + support::cpp11::to_string(-wq_info.offset));
+        build_opts.add_option("-DOUTPUT_OFFSET=" + support::cpp11::to_string(oq_info.offset));
+        build_opts.add_option("-DOUTPUT_MULTIPLIER=" + support::cpp11::to_string(output_multiplier));
+        build_opts.add_option("-DOUTPUT_SHIFT=" + support::cpp11::to_string(output_shift));
+
+        if(dwc_info.activation_info.enabled())
+        {
+            const int a_val = quantize_qasymm8(dwc_info.activation_info.a(), oq_info);
+            const int b_val = quantize_qasymm8(dwc_info.activation_info.b(), oq_info);
+            const int o1    = oq_info.offset;
+
+            build_opts.add_option("-DA_VAL=" + support::cpp11::to_string(a_val));
+            build_opts.add_option("-DB_VAL=" + support::cpp11::to_string(b_val));
+            build_opts.add_option("-DCONST_0=" + support::cpp11::to_string(o1));
+
+            const float s1 = iq_info.scale;
+            build_opts.add_option("-DS1_VAL=" + float_to_string_with_full_precision(s1));
+            build_opts.add_option("-DO1_VAL=" + support::cpp11::to_string(o1));
+        }
+    }
+    else
+    {
+        build_opts.add_option_if(dwc_info.activation_info.enabled(), "-DA_VAL=" + float_to_string_with_full_precision(dwc_info.activation_info.a()));
+        build_opts.add_option_if(dwc_info.activation_info.enabled(), "-DB_VAL=" + float_to_string_with_full_precision(dwc_info.activation_info.b()));
+    }
 
     ICLKernel::configure_internal(win_config.second);
     _kernel = static_cast<cl::Kernel>(CLKernelLibrary::get().create_kernel(kernel_name, build_opts.options()));
diff --git a/src/core/CL/kernels/CLDepthwiseConvolutionLayerReshapeWeightsGenericKernel.cpp b/src/core/CL/kernels/CLDepthwiseConvolutionLayerReshapeWeightsGenericKernel.cpp
deleted file mode 100644
index 582c600..0000000
--- a/src/core/CL/kernels/CLDepthwiseConvolutionLayerReshapeWeightsGenericKernel.cpp
+++ /dev/null
@@ -1,147 +0,0 @@
-/*
- * Copyright (c) 2017-2019 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/CL/kernels/CLDepthwiseConvolutionLayerReshapeWeightsGenericKernel.h"
-
-#include "arm_compute/core/CL/CLHelpers.h"
-#include "arm_compute/core/CL/CLKernelLibrary.h"
-#include "arm_compute/core/CL/CLValidate.h"
-#include "arm_compute/core/CL/ICLTensor.h"
-#include "arm_compute/core/CL/OpenCL.h"
-#include "arm_compute/core/Error.h"
-#include "arm_compute/core/Helpers.h"
-#include "arm_compute/core/Types.h"
-#include "support/ToolchainSupport.h"
-
-using namespace arm_compute;
-
-namespace
-{
-Status validate_arguments(const ITensorInfo *input, const ITensorInfo *output, const ITensorInfo *biases)
-{
-    const size_t idx_w = get_data_layout_dimension_index(input->data_layout(), DataLayoutDimension::WIDTH);
-    const size_t idx_h = get_data_layout_dimension_index(input->data_layout(), DataLayoutDimension::HEIGHT);
-    const size_t idx_c = get_data_layout_dimension_index(input->data_layout(), DataLayoutDimension::CHANNEL);
-
-    ARM_COMPUTE_RETURN_ERROR_ON_F16_UNSUPPORTED(input);
-    ARM_COMPUTE_RETURN_ERROR_ON_DATA_TYPE_CHANNEL_NOT_IN(input, 1, DataType::QASYMM8, DataType::F16, DataType::F32);
-    ARM_COMPUTE_RETURN_ERROR_ON_MISMATCHING_DATA_TYPES(input, output);
-    ARM_COMPUTE_RETURN_ERROR_ON(is_data_type_quantized_asymmetric(input->data_type()) && (biases != nullptr));
-    ARM_COMPUTE_RETURN_ERROR_ON(input->dimension(idx_c) != output->dimension(1));
-    ARM_COMPUTE_RETURN_ERROR_ON(output->dimension(0) != (input->dimension(idx_w) * input->dimension(idx_h) + ((biases != nullptr) ? 1 : 0)));
-    ARM_COMPUTE_RETURN_ERROR_ON_MISMATCHING_QUANTIZATION_INFO(input, output);
-
-    if(biases != nullptr)
-    {
-        ARM_COMPUTE_RETURN_ERROR_ON_MISMATCHING_DATA_TYPES(input, biases);
-        ARM_COMPUTE_RETURN_ERROR_ON(biases->dimension(0) != input->dimension(idx_c));
-        ARM_COMPUTE_RETURN_ERROR_ON(biases->num_dimensions() > 1);
-    }
-
-    return Status{};
-}
-} // namespace
-
-CLDepthwiseConvolutionLayerReshapeWeightsGenericKernel::CLDepthwiseConvolutionLayerReshapeWeightsGenericKernel()
-    : _input(nullptr), _biases(nullptr), _output(nullptr)
-{
-}
-
-void CLDepthwiseConvolutionLayerReshapeWeightsGenericKernel::configure(const ICLTensor *input, ICLTensor *output, const ICLTensor *biases)
-{
-    ARM_COMPUTE_ERROR_ON_NULLPTR(input, output);
-    ARM_COMPUTE_ERROR_THROW_ON(validate_arguments(input->info(), output->info(), (biases != nullptr) ? biases->info() : nullptr));
-
-    _input  = input;
-    _biases = biases;
-    _output = output;
-
-    const size_t idx_w = get_data_layout_dimension_index(input->info()->data_layout(), DataLayoutDimension::WIDTH);
-
-    // Create kernel
-    std::set<std::string> build_opts;
-
-    build_opts.emplace("-DDATA_TYPE=" + get_cl_type_from_data_type(input->info()->data_type()));
-    build_opts.emplace("-DSRC_WIDTH=" + support::cpp11::to_string(input->info()->dimension(idx_w)));
-    build_opts.emplace("-D" + string_from_data_layout(input->info()->data_layout()));
-    if(_biases != nullptr)
-    {
-        build_opts.emplace("-DHAS_BIAS");
-    }
-
-    _kernel = static_cast<cl::Kernel>(CLKernelLibrary::get().create_kernel("depthwise_convolution_reshape_weights_generic", build_opts));
-
-    // Configure  kernel window
-    Window win = calculate_max_window(*input->info(), Steps());
-    // The CLDepthwiseConvolutionLayerReshapeWeightsGenericKernel doesn't need padding so update_window_and_padding() can be skipped
-    output->info()->set_valid_region(ValidRegion(Coordinates(), output->info()->tensor_shape()));
-
-    ICLKernel::configure_internal(win);
-}
-
-Status CLDepthwiseConvolutionLayerReshapeWeightsGenericKernel::validate(const ITensorInfo *input, const ITensorInfo *output, const ITensorInfo *biases)
-{
-    ARM_COMPUTE_RETURN_ON_ERROR(validate_arguments(input, output, biases));
-    return Status{};
-}
-
-void CLDepthwiseConvolutionLayerReshapeWeightsGenericKernel::run(const Window &window, cl::CommandQueue &queue)
-{
-    ARM_COMPUTE_ERROR_ON_UNCONFIGURED_KERNEL(this);
-    ARM_COMPUTE_ERROR_ON_MISMATCHING_WINDOWS(ICLKernel::window(), window);
-
-    Window slice     = window.first_slice_window_3D();
-    Window slice_out = window.first_slice_window_2D();
-
-    const size_t idx_w = get_data_layout_dimension_index(_input->info()->data_layout(), DataLayoutDimension::WIDTH);
-    const size_t idx_h = get_data_layout_dimension_index(_input->info()->data_layout(), DataLayoutDimension::HEIGHT);
-    const size_t idx_c = get_data_layout_dimension_index(_input->info()->data_layout(), DataLayoutDimension::CHANNEL);
-
-    // Setup slice
-    slice.set(Window::DimX, Window::Dimension(0, _input->info()->dimension(idx_w), _input->info()->dimension(idx_w)));
-    slice.set(Window::DimY, Window::Dimension(0, _input->info()->dimension(idx_h), 1));
-    slice.set(Window::DimZ, Window::Dimension(0, _input->info()->dimension(idx_c), 1));
-
-    // Setup output slice
-    // The first two dimensions of the output are increased by the inner loops
-    slice_out.set(Window::DimX, Window::Dimension(0, 0, 0));
-    slice_out.set(Window::DimY, Window::Dimension(0, 0, 0));
-
-    // Set biases
-    if(_biases != nullptr)
-    {
-        unsigned int idx = num_arguments_per_3D_tensor() + num_arguments_per_2D_tensor();
-        Window       slice_biases;
-        slice_biases.use_tensor_dimensions(_biases->info()->tensor_shape());
-        add_1D_tensor_argument(idx, _biases, slice_biases);
-    }
-
-    do
-    {
-        unsigned int idx = 0;
-        add_3D_tensor_argument(idx, _input, slice);
-        add_2D_tensor_argument(idx, _output, slice_out);
-        enqueue(queue, *this, slice, lws_hint());
-    }
-    while(window.slide_window_slice_3D(slice) && window.slide_window_slice_2D(slice_out));
-}
diff --git a/src/core/CL/kernels/CLDepthwiseIm2ColKernel.cpp b/src/core/CL/kernels/CLDepthwiseIm2ColKernel.cpp
deleted file mode 100644
index 0312a57..0000000
--- a/src/core/CL/kernels/CLDepthwiseIm2ColKernel.cpp
+++ /dev/null
@@ -1,148 +0,0 @@
-/*
- * Copyright (c) 2017-2019 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/CL/kernels/CLDepthwiseIm2ColKernel.h"
-
-#include "arm_compute/core/CL/CLHelpers.h"
-#include "arm_compute/core/CL/CLKernelLibrary.h"
-#include "arm_compute/core/CL/CLValidate.h"
-#include "arm_compute/core/CL/ICLTensor.h"
-#include "arm_compute/core/CL/OpenCL.h"
-#include "arm_compute/core/Error.h"
-#include "arm_compute/core/Helpers.h"
-#include "arm_compute/core/Types.h"
-#include "support/ToolchainSupport.h"
-
-#include <tuple>
-
-using namespace arm_compute;
-
-CLDepthwiseIm2ColKernel::CLDepthwiseIm2ColKernel()
-    : _input(nullptr), _output(nullptr)
-{
-}
-
-namespace
-{
-Status validate_arguments(const ITensorInfo *input, const ITensorInfo *output, const Size2D &kernel_dims, const PadStrideInfo &conv_info, bool has_bias, unsigned int depth_multiplier,
-                          const Size2D &dilation)
-{
-    const size_t idx_c = get_data_layout_dimension_index(input->data_layout(), DataLayoutDimension::CHANNEL);
-
-    ARM_COMPUTE_UNUSED(conv_info);
-    ARM_COMPUTE_RETURN_ERROR_ON_F16_UNSUPPORTED(input);
-    ARM_COMPUTE_RETURN_ERROR_ON_DATA_TYPE_CHANNEL_NOT_IN(input, 1, DataType::QASYMM8, DataType::F16, DataType::F32);
-    ARM_COMPUTE_RETURN_ERROR_ON_MISMATCHING_DATA_TYPES(input, output);
-    ARM_COMPUTE_RETURN_ERROR_ON(is_data_type_quantized_asymmetric(input->data_type()) && has_bias);
-    ARM_COMPUTE_RETURN_ERROR_ON((input->dimension(idx_c) * depth_multiplier) != output->dimension(2));
-    ARM_COMPUTE_RETURN_ERROR_ON(output->dimension(0) != (kernel_dims.width * kernel_dims.height + ((has_bias) ? 1 : 0)));
-    ARM_COMPUTE_RETURN_ERROR_ON((dilation.x() < 1) || dilation.y() < 1);
-    ARM_COMPUTE_RETURN_ERROR_ON_MISMATCHING_QUANTIZATION_INFO(input, output);
-
-    return Status{};
-}
-} // namespace
-
-void CLDepthwiseIm2ColKernel::configure(const ICLTensor *input, ICLTensor *output, const Size2D &kernel_dims, const PadStrideInfo &conv_info, bool has_bias, unsigned int depth_multiplier,
-                                        const Size2D &dilation)
-{
-    ARM_COMPUTE_ERROR_ON_NULLPTR(input, output);
-    ARM_COMPUTE_ERROR_THROW_ON(validate_arguments(input->info(), output->info(), kernel_dims, conv_info, has_bias, depth_multiplier, dilation));
-
-    _input  = input;
-    _output = output;
-
-    const DataLayout              data_layout = input->info()->data_layout();
-    const size_t                  idx_w       = get_data_layout_dimension_index(data_layout, DataLayoutDimension::WIDTH);
-    const size_t                  idx_h       = get_data_layout_dimension_index(data_layout, DataLayoutDimension::HEIGHT);
-    const UniformQuantizationInfo qinfo       = input->info()->quantization_info().uniform();
-
-    // Create kernel
-    CLBuildOptions build_opts;
-
-    build_opts.add_option("-DDATA_TYPE=" + get_cl_type_from_data_type(input->info()->data_type()));
-    build_opts.add_option("-DSTRIDE_X=" + support::cpp11::to_string(conv_info.stride().first));
-    build_opts.add_option("-DSTRIDE_Y=" + support::cpp11::to_string(conv_info.stride().second));
-    build_opts.add_option("-DPAD_LEFT=" + support::cpp11::to_string(conv_info.pad_left()));
-    build_opts.add_option("-DPAD_TOP=" + support::cpp11::to_string(conv_info.pad_top()));
-    build_opts.add_option("-DPAD_RIGHT=" + support::cpp11::to_string(conv_info.pad_right()));
-    build_opts.add_option("-DPAD_BOTTOM=" + support::cpp11::to_string(conv_info.pad_bottom()));
-    build_opts.add_option("-DSRC_WIDTH=" + support::cpp11::to_string(input->info()->dimension(idx_w)));
-    build_opts.add_option("-DSRC_HEIGHT=" + support::cpp11::to_string(input->info()->dimension(idx_h)));
-    build_opts.add_option("-DKERNEL_WIDTH=" + support::cpp11::to_string(kernel_dims.width));
-    build_opts.add_option("-DKERNEL_HEIGHT=" + support::cpp11::to_string(kernel_dims.height));
-    build_opts.add_option("-DDEPTH_MULTIPLIER=" + support::cpp11::to_string(depth_multiplier));
-    build_opts.add_option("-DDILATION_X=" + support::cpp11::to_string(dilation.x()));
-    build_opts.add_option("-DDILATION_Y=" + support::cpp11::to_string(dilation.y()));
-    build_opts.add_option("-D" + string_from_data_layout(input->info()->data_layout()));
-    build_opts.add_option_if(has_bias, "-DHAS_BIAS");
-    build_opts.add_option_if_else(is_data_type_quantized_asymmetric(input->info()->data_type()),
-                                  "-DPAD_VALUE=" + support::cpp11::to_string(qinfo.offset),
-                                  "-DPAD_VALUE=0");
-
-    _kernel = static_cast<cl::Kernel>(CLKernelLibrary::get().create_kernel("depthwise_im2col", build_opts.options()));
-
-    // Configure  kernel window
-    Window win = calculate_max_window(*output->info(), Steps());
-    // CLDepthwiseIm2ColKernel doesn't need padding so update_window_and_padding() can be skipped
-    output->info()->set_valid_region(ValidRegion(Coordinates(), output->info()->tensor_shape()));
-
-    ICLKernel::configure_internal(win);
-}
-
-Status CLDepthwiseIm2ColKernel::validate(const ITensorInfo *input, const ITensorInfo *output, const Size2D &kernel_dims, const PadStrideInfo &conv_info, bool has_bias, unsigned int depth_multiplier,
-                                         const Size2D &dilation)
-{
-    ARM_COMPUTE_RETURN_ON_ERROR(validate_arguments(input, output, kernel_dims, conv_info, has_bias, depth_multiplier, dilation));
-
-    return Status{};
-}
-
-void CLDepthwiseIm2ColKernel::run(const Window &window, cl::CommandQueue &queue)
-{
-    ARM_COMPUTE_ERROR_ON_UNCONFIGURED_KERNEL(this);
-    ARM_COMPUTE_ERROR_ON_MISMATCHING_WINDOWS(ICLKernel::window(), window);
-
-    Window slice    = window.first_slice_window_3D();
-    Window slice_in = window.first_slice_window_3D();
-
-    // Setup slice
-    slice.set(Window::DimX, Window::Dimension(0, _output->info()->dimension(0), _output->info()->dimension(0)));
-    slice.set(Window::DimY, Window::Dimension(0, _output->info()->dimension(1), 1));
-    slice.set(Window::DimZ, Window::Dimension(0, _output->info()->dimension(2), 1));
-
-    // Setup input slice
-    // The first three dimensions of the input are increased by the inner loops
-    slice_in.set(Window::DimX, Window::Dimension(0, 0, 0));
-    slice_in.set(Window::DimY, Window::Dimension(0, 0, 0));
-    slice_in.set(Window::DimZ, Window::Dimension(0, 0, 0));
-
-    do
-    {
-        unsigned int idx = 0;
-        add_3D_tensor_argument(idx, _input, slice_in);
-        add_3D_tensor_argument(idx, _output, slice);
-        enqueue(queue, *this, slice, lws_hint());
-    }
-    while(window.slide_window_slice_3D(slice) && window.slide_window_slice_3D(slice_in));
-}
diff --git a/src/core/CL/kernels/CLDepthwiseVectorToTensorKernel.cpp b/src/core/CL/kernels/CLDepthwiseVectorToTensorKernel.cpp
deleted file mode 100644
index 0f029fd..0000000
--- a/src/core/CL/kernels/CLDepthwiseVectorToTensorKernel.cpp
+++ /dev/null
@@ -1,125 +0,0 @@
-/*
- * Copyright (c) 2017-2019 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/CL/kernels/CLDepthwiseVectorToTensorKernel.h"
-
-#include "arm_compute/core/CL/CLHelpers.h"
-#include "arm_compute/core/CL/CLKernelLibrary.h"
-#include "arm_compute/core/CL/CLValidate.h"
-#include "arm_compute/core/CL/ICLTensor.h"
-#include "arm_compute/core/CL/OpenCL.h"
-#include "arm_compute/core/Error.h"
-#include "arm_compute/core/Helpers.h"
-#include "arm_compute/core/Types.h"
-#include "arm_compute/core/utils/misc/ShapeCalculator.h"
-#include "support/ToolchainSupport.h"
-
-using namespace arm_compute;
-using namespace arm_compute::misc::shape_calculator;
-
-namespace
-{
-Status validate_arguments(const ITensorInfo *input, const ITensorInfo *output, size_t conv_w, size_t conv_h)
-{
-    ARM_COMPUTE_RETURN_ERROR_ON_F16_UNSUPPORTED(input);
-    ARM_COMPUTE_RETURN_ERROR_ON_DATA_TYPE_CHANNEL_NOT_IN(input, 1, DataType::QASYMM8, DataType::S32, DataType::F16, DataType::F32);
-
-    if(output->total_size() != 0)
-    {
-        TensorShape output_shape = compute_vector_to_tensor_output_shape(input->tensor_shape(), conv_w, conv_h, output->data_layout());
-        ARM_COMPUTE_RETURN_ERROR_ON_MISMATCHING_DIMENSIONS(output->tensor_shape(), output_shape);
-        ARM_COMPUTE_RETURN_ERROR_ON_MISMATCHING_DATA_TYPES(input, output);
-        ARM_COMPUTE_RETURN_ERROR_ON_MISMATCHING_QUANTIZATION_INFO(input, output);
-    }
-
-    return Status{};
-}
-} // namespace
-
-CLDepthwiseVectorToTensorKernel::CLDepthwiseVectorToTensorKernel()
-    : _input(nullptr), _output(nullptr)
-{
-}
-
-void CLDepthwiseVectorToTensorKernel::configure(const ICLTensor *input, ICLTensor *output, size_t conv_w, size_t conv_h)
-{
-    ARM_COMPUTE_ERROR_ON_NULLPTR(input, output);
-
-    // Output auto inizialitation if not yet initialized
-    TensorShape output_shape = compute_vector_to_tensor_output_shape(input->info()->tensor_shape(), conv_w, conv_h, output->info()->data_layout());
-    auto_init_if_empty(*output->info(), input->info()->clone()->set_tensor_shape(output_shape));
-
-    ARM_COMPUTE_ERROR_THROW_ON(validate_arguments(input->info(), output->info(), conv_w, conv_h));
-
-    _input  = input;
-    _output = output;
-
-    // Create kernel
-    CLBuildOptions build_opts;
-    build_opts.add_option("-DDATA_TYPE=" + get_cl_type_from_data_type(input->info()->data_type()));
-    build_opts.add_option("-DCONV_WIDTH=" + support::cpp11::to_string(conv_w));
-    build_opts.add_option("-DCONV_HEIGHT=" + support::cpp11::to_string(conv_h));
-    build_opts.add_option("-D" + string_from_data_layout(output->info()->data_layout()));
-
-    _kernel = static_cast<cl::Kernel>(CLKernelLibrary::get().create_kernel("depthwise_vector_to_tensor", build_opts.options()));
-
-    // Configure  kernel window
-    Window win = calculate_max_window(*input->info(), Steps());
-    // The CLDepthwisevectorToTensorKernel doesn't need padding so update_window_and_padding() can be skipped
-    output->info()->set_valid_region(ValidRegion(Coordinates(), output->info()->tensor_shape()));
-
-    ICLKernel::configure_internal(win);
-}
-
-Status CLDepthwiseVectorToTensorKernel::validate(const ITensorInfo *input, const ITensorInfo *output, size_t conv_w, size_t conv_h)
-{
-    ARM_COMPUTE_RETURN_ON_ERROR(validate_arguments(input, output, conv_w, conv_h));
-    return Status{};
-}
-
-void CLDepthwiseVectorToTensorKernel::run(const Window &window, cl::CommandQueue &queue)
-{
-    ARM_COMPUTE_ERROR_ON_UNCONFIGURED_KERNEL(this);
-    ARM_COMPUTE_ERROR_ON_MISMATCHING_WINDOWS(ICLKernel::window(), window);
-
-    Window slice     = window.first_slice_window_1D();
-    Window slice_out = window.first_slice_window_3D();
-
-    // Setup slice
-    slice.set(Window::DimX, Window::Dimension(0, _input->info()->dimension(0), 1));
-
-    // Setup output slice
-    // The first three dimensions of the output are increased by the inner loops
-    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));
-
-    do
-    {
-        unsigned int idx = 0;
-        add_1D_tensor_argument(idx, _input, slice);
-        add_3D_tensor_argument(idx, _output, slice_out);
-        enqueue(queue, *this, slice, lws_hint());
-    }
-    while(window.slide_window_slice_1D(slice) && window.slide_window_slice_3D(slice_out));
-}
diff --git a/src/core/CL/kernels/CLDirectConvolutionOutputStageKernel.cpp b/src/core/CL/kernels/CLDirectConvolutionOutputStageKernel.cpp
deleted file mode 100644
index 515321c..0000000
--- a/src/core/CL/kernels/CLDirectConvolutionOutputStageKernel.cpp
+++ /dev/null
@@ -1,209 +0,0 @@
-/*
- * Copyright (c) 2018-2019 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/CL/kernels/CLDirectConvolutionLayerOutputStageKernel.h"
-
-#include "arm_compute/core/AccessWindowStatic.h"
-#include "arm_compute/core/CL/CLValidate.h"
-#include "arm_compute/core/CL/ICLTensor.h"
-#include "arm_compute/core/Error.h"
-#include "arm_compute/core/Helpers.h"
-#include "arm_compute/core/Types.h"
-#include "arm_compute/core/Window.h"
-
-#include <cstddef>
-#include <cstdint>
-
-namespace arm_compute
-{
-namespace
-{
-Status validate_arguments(const ITensorInfo *input, const ITensorInfo *bias, const ITensorInfo *output)
-{
-    ARM_COMPUTE_RETURN_ERROR_ON_NULLPTR(input);
-    ARM_COMPUTE_RETURN_ERROR_ON_F16_UNSUPPORTED(input);
-    ARM_COMPUTE_RETURN_ERROR_ON_DATA_TYPE_CHANNEL_NOT_IN(input, 1, DataType::S32, DataType::F16,
-                                                         DataType::F32);
-
-    if(bias != nullptr)
-    {
-        ARM_COMPUTE_RETURN_ERROR_ON_F16_UNSUPPORTED(bias);
-        ARM_COMPUTE_RETURN_ERROR_ON_DATA_TYPE_CHANNEL_NOT_IN(bias, 1, DataType::S32, DataType::F16, DataType::F32);
-
-        if(is_data_type_quantized_asymmetric(input->data_type()))
-        {
-            ARM_COMPUTE_RETURN_ERROR_ON_DATA_TYPE_CHANNEL_NOT_IN(bias, 1, DataType::S32);
-        }
-        else
-        {
-            ARM_COMPUTE_RETURN_ERROR_ON_MISMATCHING_DATA_TYPES(input, bias);
-        }
-
-        ARM_COMPUTE_RETURN_ERROR_ON(bias->num_dimensions() > 1);
-    }
-    else
-    {
-        ARM_COMPUTE_RETURN_ERROR_ON_MSG(is_data_type_float(input->data_type()),
-                                        "Calling output stage kernel with floating point arguments");
-    }
-
-    // Checks performed on output
-    if(input->data_type() == DataType::S32)
-    {
-        // Quantized configuration checks
-        ARM_COMPUTE_RETURN_ERROR_ON_DATA_TYPE_CHANNEL_NOT_IN(output, 1, DataType::QASYMM8);
-    }
-    else
-    {
-        // In case of out-of-place computation (supported for non-quantized configurations)
-        if((output != nullptr) && (output->total_size() != 0))
-        {
-            ARM_COMPUTE_RETURN_ERROR_ON_MISMATCHING_DATA_TYPES(input, output);
-        }
-    }
-
-    return Status{};
-}
-
-std::pair<Status, Window> validate_and_configure_window(ITensorInfo *input, ITensorInfo *bias, ITensorInfo *output)
-{
-    bool               window_changed                    = false;
-    const unsigned int num_elems_processed_per_iteration = 16 / element_size_from_data_type(input->data_type());
-
-    // Configure kernel window
-    Window win = calculate_max_window(*input, Steps(num_elems_processed_per_iteration));
-
-    // Input window
-    AccessWindowHorizontal input_access(input, 0, num_elems_processed_per_iteration);
-    window_changed = window_changed || update_window_and_padding(win, input_access);
-
-    // Bias window
-    if(bias != nullptr)
-    {
-        AccessWindowStatic bias_access(bias, 0, 0, ceil_to_multiple(bias->dimension(0), num_elems_processed_per_iteration), bias->dimension(1));
-        window_changed = window_changed || update_window_and_padding(win, bias_access);
-    }
-
-    // Output window
-    if(output != nullptr && (output->total_size() != 0))
-    {
-        AccessWindowHorizontal output_access(output, 0, num_elems_processed_per_iteration);
-        window_changed = window_changed || update_window_and_padding(win, output_access);
-        output_access.set_valid_region(win, ValidRegion(Coordinates(), output->tensor_shape()));
-    }
-    else
-    {
-        input_access.set_valid_region(win, ValidRegion(Coordinates(), input->tensor_shape()));
-    }
-
-    Status err = (window_changed) ? ARM_COMPUTE_CREATE_ERROR(ErrorCode::RUNTIME_ERROR, "Insufficient Padding!") : Status{};
-    return std::make_pair(err, win);
-}
-} // namespace
-
-CLDirectConvolutionLayerOutputStageKernel::CLDirectConvolutionLayerOutputStageKernel()
-    : _input(nullptr), _bias(nullptr), _output(nullptr), _result_fixedpoint_multiplier(0), _result_shift(0), _result_offset_after_shift(0)
-{
-}
-
-void CLDirectConvolutionLayerOutputStageKernel::configure(ICLTensor *input, const ICLTensor *bias, ICLTensor *output,
-                                                          int result_fixedpoint_multiplier, int result_shift, int result_offset_after_shift)
-{
-    ARM_COMPUTE_ERROR_ON_NULLPTR(input);
-
-    // Auto-initialize output if required
-    if(output != nullptr)
-    {
-        // Work out expected output data type
-        const DataType output_dt = (input->info()->data_type() == DataType::S32) ? DataType::QASYMM8 : input->info()->data_type();
-        // Output tensor auto initialization if not yet initialized
-        auto_init_if_empty(*output->info(), input->info()->clone()->set_data_type(output_dt));
-    }
-
-    // Perform validation step
-    ARM_COMPUTE_ERROR_THROW_ON(validate_arguments(input->info(), (bias == nullptr) ? nullptr : bias->info(), (output == nullptr) ? nullptr : output->info()));
-
-    _bias                         = bias;
-    _input                        = input;
-    _output                       = output;
-    _result_fixedpoint_multiplier = result_fixedpoint_multiplier;
-    _result_shift                 = result_shift;
-    _result_offset_after_shift    = result_offset_after_shift;
-
-    const unsigned int num_elems_accessed_per_iteration = 16 / element_size_from_data_type(input->info()->data_type());
-
-    // Create kernel
-    CLBuildOptions build_opts;
-    build_opts.add_option_if(bias != nullptr, "-DHAS_BIAS");
-    build_opts.add_option("-D" + string_from_data_layout(input->info()->data_layout()));
-    build_opts.add_option("-DVEC_SIZE=" + support::cpp11::to_string(num_elems_accessed_per_iteration));
-    _kernel = static_cast<cl::Kernel>(CLKernelLibrary::get().create_kernel("output_stage_quantized", build_opts.options()));
-
-    // Set static kernel arguments
-    int idx = 2 * num_arguments_per_3D_tensor() + ((bias != nullptr) ? num_arguments_per_1D_tensor() : 0);
-    _kernel.setArg<int>(idx++, _result_offset_after_shift);
-    _kernel.setArg<int>(idx++, _result_fixedpoint_multiplier);
-    _kernel.setArg<int>(idx++, _result_shift);
-
-    // Configure kernel window
-    auto win_config = validate_and_configure_window(input->info(), (bias == nullptr) ? nullptr : bias->info(), (output == nullptr) ? nullptr : output->info());
-    ARM_COMPUTE_ERROR_THROW_ON(win_config.first);
-    ICLKernel::configure_internal(win_config.second);
-}
-
-Status CLDirectConvolutionLayerOutputStageKernel::validate(const ITensorInfo *input, const ITensorInfo *bias, const ITensorInfo *output)
-{
-    ARM_COMPUTE_RETURN_ON_ERROR(validate_arguments(input, bias, output));
-    ARM_COMPUTE_RETURN_ON_ERROR(validate_and_configure_window(input->clone().get(), bias == nullptr ? nullptr : bias->clone().get(), output == nullptr ? nullptr : output->clone().get()).first);
-
-    return Status{};
-}
-
-void CLDirectConvolutionLayerOutputStageKernel::run(const Window &window, cl::CommandQueue &queue)
-{
-    ARM_COMPUTE_ERROR_ON_UNCONFIGURED_KERNEL(this);
-    ARM_COMPUTE_ERROR_ON_MISMATCHING_WINDOWS(ICLKernel::window(), window);
-
-    Window slice = window.first_slice_window_3D();
-
-    // Set bias vector
-    if(_bias != nullptr)
-    {
-        unsigned int idx1 = 2 * num_arguments_per_3D_tensor();
-        Window       slice_biases;
-        slice_biases.use_tensor_dimensions(_bias->info()->tensor_shape());
-        add_1D_tensor_argument(idx1, _bias, slice_biases);
-    }
-
-    // Run kernel
-    do
-    {
-        // Set arguments
-        unsigned int idx = 0;
-        add_3D_tensor_argument(idx, _input, slice);
-        add_3D_tensor_argument(idx, _output, slice);
-        enqueue(queue, *this, slice, lws_hint());
-    }
-    while(window.slide_window_slice_3D(slice));
-}
-} // namespace arm_compute
diff --git a/src/core/CL/kernels/CLGEMMLowpQuantizeDownInt32ToUint8ScaleByFloatKernel.cpp b/src/core/CL/kernels/CLGEMMLowpQuantizeDownInt32ToUint8ScaleByFloatKernel.cpp
index 2967a73..ae096f2 100644
--- a/src/core/CL/kernels/CLGEMMLowpQuantizeDownInt32ToUint8ScaleByFloatKernel.cpp
+++ b/src/core/CL/kernels/CLGEMMLowpQuantizeDownInt32ToUint8ScaleByFloatKernel.cpp
@@ -35,8 +35,6 @@
 
 #include "support/ToolchainSupport.h"
 
-using namespace arm_compute;
-
 namespace arm_compute
 {
 namespace
@@ -101,8 +99,6 @@
 } // namespace
 
 class Coordinates;
-} // namespace arm_compute
-
 CLGEMMLowpQuantizeDownInt32ToUint8ScaleByFloatKernel::CLGEMMLowpQuantizeDownInt32ToUint8ScaleByFloatKernel()
     : _input(nullptr), _bias(nullptr), _output(nullptr)
 {
@@ -177,3 +173,4 @@
     }
     while(collapsed.slide_window_slice_3D(slice));
 }
+} // namespace arm_compute