Files
opencv-MIRROR/modules/dnn/src/layers/reshape_layer.cpp
Yashas Samaga B L 613c12e590 Merge pull request #14827 from YashasSamaga:cuda4dnn-csl-low
CUDA backend for the DNN module

* stub cuda4dnn design

* minor fixes for tests and doxygen

* add csl public api directory to module headers

* add low-level CSL components

* add high-level CSL components

* integrate csl::Tensor into backbone code

* switch to CPU iff unsupported; otherwise, fail on error

* add fully connected layer

* add softmax layer

* add activation layers

* support arbitary rank TensorDescriptor

* pass input wrappers to `initCUDA()`

* add 1d/2d/3d-convolution

* add pooling layer

* reorganize and refactor code

* fixes for gcc, clang and doxygen; remove cxx14/17 code

* add blank_layer

* add LRN layer

* add rounding modes for pooling layer

* split tensor.hpp into tensor.hpp and tensor_ops.hpp

* add concat layer

* add scale layer

* add batch normalization layer

* split math.cu into activations.cu and math.hpp

* add eltwise layer

* add flatten layer

* add tensor transform api

* add asymmetric padding support for convolution layer

* add reshape layer

* fix rebase issues

* add permute layer

* add padding support for concat layer

* refactor and reorganize code

* add normalize layer

* optimize bias addition in scale layer

* add prior box layer

* fix and optimize normalize layer

* add asymmetric padding support for pooling layer

* add event API

* improve pooling performance for some padding scenarios

* avoid over-allocation of compute resources to kernels

* improve prior box performance

* enable layer fusion

* add const layer

* add resize layer

* add slice layer

* add padding layer

* add deconvolution layer

* fix channelwise  ReLU initialization

* add vector traits

* add vectorized versions of relu, clipped_relu, power

* add vectorized concat kernels

* improve concat_with_offsets performance

* vectorize scale and bias kernels

* add support for multi-billion element tensors

* vectorize prior box kernels

* fix address alignment check

* improve bias addition performance of conv/deconv/fc layers

* restructure code for supporting multiple targets

* add DNN_TARGET_CUDA_FP64

* add DNN_TARGET_FP16

* improve vectorization

* add region layer

* improve tensor API, add dynamic ranks

1. use ManagedPtr instead of a Tensor in backend wrapper
2. add new methods to tensor classes
  - size_range: computes the combined size of for a given axis range
  - tensor span/view can be constructed from a raw pointer and shape
3. the tensor classes can change their rank at runtime (previously rank was fixed at compile-time)
4. remove device code from tensor classes (as they are unused)
5. enforce strict conditions on tensor class APIs to improve debugging ability

* fix parametric relu activation

* add squeeze/unsqueeze tensor API

* add reorg layer

* optimize permute and enable 2d permute

* enable 1d and 2d slice

* add split layer

* add shuffle channel layer

* allow tensors of different ranks in reshape primitive

* patch SliceOp to allow Crop Layer

* allow extra shape inputs in reshape layer

* use `std::move_backward` instead of `std::move` for insert in resizable_static_array

* improve workspace management

* add spatial LRN

* add nms (cpu) to region layer

* add max pooling with argmax ( and a fix to limits.hpp)

* add max unpooling layer

* rename DNN_TARGET_CUDA_FP32 to DNN_TARGET_CUDA

* update supportBackend to be more rigorous

* remove stray include from preventing non-cuda build

* include op_cuda.hpp outside condition #if

* refactoring, fixes and many optimizations

* drop DNN_TARGET_CUDA_FP64

* fix gcc errors

* increase max. tensor rank limit to six

* add Interp layer

* drop custom layers; use BackendNode

* vectorize activation kernels

* fixes for gcc

* remove wrong assertion

* fix broken assertion in unpooling primitive

* fix build errors in non-CUDA build

* completely remove workspace from public API

* fix permute layer

* enable accuracy and perf. tests for DNN_TARGET_CUDA

* add asynchronous forward

* vectorize eltwise ops

* vectorize fill kernel

* fixes for gcc

* remove CSL headers from public API

* remove csl header source group from cmake

* update min. cudnn version in cmake

* add numerically stable FP32 log1pexp

* refactor code

* add FP16 specialization to cudnn based tensor addition

* vectorize scale1 and bias1 + minor refactoring

* fix doxygen build

* fix invalid alignment assertion

* clear backend wrappers before allocateLayers

* ignore memory lock failures

* do not allocate internal blobs

* integrate NVTX

* add numerically stable half precision log1pexp

* fix indentation, following coding style,  improve docs

* remove accidental modification of IE code

* Revert "add asynchronous forward"

This reverts commit 1154b9da9da07e9b52f8a81bdcea48cf31c56f70.

* [cmake] throw error for unsupported CC versions

* fix rebase issues

* add more docs, refactor code, fix bugs

* minor refactoring and fixes

* resolve warnings/errors from clang

* remove haveCUDA() checks from supportBackend()

* remove NVTX integration

* changes based on review comments

* avoid exception when no CUDA device is present

* add color code for CUDA in Net::dump
2019-10-21 14:28:00 +03:00

302 lines
10 KiB
C++

/*M///////////////////////////////////////////////////////////////////////////////////////
//
// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
//
// By downloading, copying, installing or using the software you agree to this license.
// If you do not agree to this license, do not download, install,
// copy or use the software.
//
//
// License Agreement
// For Open Source Computer Vision Library
//
// Copyright (C) 2013, OpenCV Foundation, all rights reserved.
// Copyright (C) 2017, Intel Corporation, all rights reserved.
// Third party copyrights are property of their respective owners.
//
// Redistribution and use in source and binary forms, with or without modification,
// are permitted provided that the following conditions are met:
//
// * Redistribution's of source code must retain the above copyright notice,
// this list of conditions and the following disclaimer.
//
// * Redistribution's in binary form must reproduce the above copyright notice,
// this list of conditions and the following disclaimer in the documentation
// and/or other materials provided with the distribution.
//
// * The name of the copyright holders may not be used to endorse or promote products
// derived from this software without specific prior written permission.
//
// This software is provided by the copyright holders and contributors "as is" and
// any express or implied warranties, including, but not limited to, the implied
// warranties of merchantability and fitness for a particular purpose are disclaimed.
// In no event shall the Intel Corporation or contributors be liable for any direct,
// indirect, incidental, special, exemplary, or consequential damages
// (including, but not limited to, procurement of substitute goods or services;
// loss of use, data, or profits; or business interruption) however caused
// and on any theory of liability, whether in contract, strict liability,
// or tort (including negligence or otherwise) arising in any way out of
// the use of this software, even if advised of the possibility of such damage.
//
//M*/
#include "../precomp.hpp"
#include "layers_common.hpp"
#include "../op_cuda.hpp"
#include "../op_inf_engine.hpp"
#include <opencv2/dnn/shape_utils.hpp>
#ifdef HAVE_CUDA
#include "../cuda4dnn/primitives/reshape.hpp"
using namespace cv::dnn::cuda4dnn;
#endif
namespace cv
{
namespace dnn
{
static void computeShapeByReshapeMask(const MatShape &srcShape,
const MatShape &maskShape,
Range srcRange /*= Range::all()*/,
MatShape& dstShape)
{
int srcShapeSize = (int)srcShape.size();
int maskShapeSize = (int)maskShape.size();
if (srcRange == Range::all())
srcRange = Range(0, srcShapeSize);
else
{
int sz = srcRange.size();
srcRange.start = clamp(srcRange.start, srcShapeSize);
srcRange.end = srcRange.end == INT_MAX ? srcShapeSize : srcRange.start + sz;
}
bool explicitMask = !maskShape.empty(); // All mask values are positive.
for (int i = 0, n = maskShape.size(); i < n && explicitMask; ++i)
{
explicitMask = maskShape[i] > 0;
}
// Working range of source shape is a range where area(src) == area(mask).
if (explicitMask)
{
int maskTotal = total(maskShape);
// Go from the end of mask until we collect required total.
bool matched = false;
for (int i = srcRange.end - 1; i >= srcRange.start; --i)
{
if (matched)
{
if (total(srcShape, i, srcRange.end) != maskTotal)
{
srcRange.start = i + 1;
break;
}
else if (i == 0)
{
srcRange.start = 0;
break;
}
}
else
{
matched = total(srcShape, i, srcRange.end) == maskTotal;
}
}
while (total(srcShape, srcRange.start, srcRange.end) != maskTotal && srcRange.start > 0)
{
srcRange.start -= 1;
}
CV_Assert(total(srcShape, srcRange.start, srcRange.end) == maskTotal);
}
CV_Assert(0 <= srcRange.start && srcRange.start <= srcRange.end && srcRange.end <= srcShapeSize);
int dstShapeSize = srcShapeSize - srcRange.size() + maskShapeSize;
dstShape.resize(dstShapeSize);
std::copy(srcShape.begin(), srcShape.begin() + srcRange.start, dstShape.begin());
std::copy(srcShape.begin() + srcRange.end, srcShape.begin() + srcShapeSize, dstShape.begin() + srcRange.start + maskShapeSize);
int inferDim = -1;
for (int i = 0; i < maskShapeSize; i++)
{
if (maskShape[i] > 0)
{
dstShape[srcRange.start + i] = maskShape[i];
}
else if (maskShape[i] == 0)
{
if (srcRange.start + i >= srcShapeSize)
CV_Error(Error::StsBadArg, format("Copy dim[%d] (which has zero size) is out of the source shape bounds", srcRange.start + i));
dstShape[srcRange.start + i] = srcShape[srcRange.start + i];
}
else if (maskShape[i] == -1)
{
if (inferDim != -1)
CV_Error(Error::StsAssert, "Duplicate of inferred dim (which is denoted by -1)");
inferDim = srcRange.start + i;
dstShape[inferDim] = 1;
}
else
CV_Error(Error::StsBadArg, "maskShape[i] >= -1");
}
size_t srcTotal = total(srcShape);
size_t dstTotal = total(dstShape);
CV_Assert(dstTotal != 0);
if (inferDim != -1)
{
if (srcTotal % dstTotal != 0)
CV_Error(Error::StsBackTrace, "Can't infer a dim denoted by -1");
dstShape[inferDim] = (int)(srcTotal / dstTotal);
}
else
{
CV_Assert(srcTotal == dstTotal);
}
}
class ReshapeLayerImpl CV_FINAL : public ReshapeLayer
{
public:
ReshapeLayerImpl(const LayerParams& params)
{
setParamsFrom(params);
int axis = params.get<int>("axis", 0);
int numAxes = params.get<int>("num_axes", -1);
CV_Assert(numAxes >= -1);
newShapeRange = (numAxes == -1) ? Range(axis, INT_MAX) : Range(axis, axis + numAxes);
newShapeDesc.clear();
if (params.has("dim"))
{
const DictValue &paramShape = params.get("dim");
int i, dims = paramShape.size();
newShapeDesc.resize(dims);
for (i = 0; i < dims; i++)
newShapeDesc[i] = paramShape.get<int>(i);
}
}
virtual bool supportBackend(int backendId) CV_OVERRIDE
{
return backendId == DNN_BACKEND_OPENCV ||
backendId == DNN_BACKEND_CUDA ||
(backendId == DNN_BACKEND_INFERENCE_ENGINE && haveInfEngine());
}
bool getMemoryShapes(const std::vector<MatShape> &inputs,
const int requiredOutputs,
std::vector<MatShape> &outputs,
std::vector<MatShape> &internals) const CV_OVERRIDE
{
if (inputs.size() == 1 || inputs.size() == requiredOutputs)
{
outputs.clear();
for (size_t i = 0; i < inputs.size(); i++)
{
outputs.push_back(MatShape());
computeShapeByReshapeMask(inputs[i], newShapeDesc, newShapeRange, outputs.back());
}
}
else
{
CV_Assert_N(inputs.size() == 2, total(inputs[0]) == total(inputs[1]));
outputs.assign(1, inputs[1]);
}
return true;
}
void finalize(InputArrayOfArrays, OutputArrayOfArrays outputs_arr) CV_OVERRIDE
{
std::vector<Mat> outputs;
outputs_arr.getMatVector(outputs);
CV_Assert(!outputs.empty());
outShapes.resize(outputs.size());
for (int i = 0; i < outputs.size(); ++i)
outShapes[i] = shape(outputs[i]);
}
bool forward_ocl(InputArrayOfArrays inps, OutputArrayOfArrays outs, OutputArrayOfArrays internals)
{
std::vector<UMat> inputs;
std::vector<UMat> outputs;
inps.getUMatVector(inputs);
outs.getUMatVector(outputs);
for (size_t i = 0; i < outputs.size(); i++)
{
UMat srcBlob = inputs[i];
void *src_handle = inputs[i].handle(ACCESS_READ);
void *dst_handle = outputs[i].handle(ACCESS_WRITE);
if (src_handle != dst_handle)
{
UMat umat = srcBlob.reshape(1, (int)outShapes[i].size(), &outShapes[i][0]);
umat.copyTo(outputs[i]);
}
}
outs.assign(outputs);
return true;
}
void forward(InputArrayOfArrays inputs_arr, OutputArrayOfArrays outputs_arr, OutputArrayOfArrays internals_arr) CV_OVERRIDE
{
CV_TRACE_FUNCTION();
CV_TRACE_ARG_VALUE(name, "name", name.c_str());
CV_OCL_RUN(IS_DNN_OPENCL_TARGET(preferableTarget),
forward_ocl(inputs_arr, outputs_arr, internals_arr))
std::vector<Mat> inputs, outputs;
inputs_arr.getMatVector(inputs);
outputs_arr.getMatVector(outputs);
for (size_t i = 0; i < outputs.size(); i++)
{
Mat srcBlob = inputs[i];
if (outputs[i].data != srcBlob.data)
srcBlob.reshape(1, shape(outputs[i])).copyTo(outputs[i]);
}
}
#ifdef HAVE_CUDA
Ptr<BackendNode> initCUDA(
void *context_,
const std::vector<Ptr<BackendWrapper>>& inputs,
const std::vector<Ptr<BackendWrapper>>& outputs
) override
{
auto context = reinterpret_cast<csl::CSLContext*>(context_);
return make_cuda_node<cuda4dnn::ReshapeOp>(preferableTarget, std::move(context->stream));
}
#endif
#ifdef HAVE_INF_ENGINE
virtual Ptr<BackendNode> initInfEngine(const std::vector<Ptr<BackendWrapper> >& inputs) CV_OVERRIDE
{
InferenceEngine::Builder::ReshapeLayer ieLayer(name);
CV_Assert(outShapes.size() == 1);
ieLayer.setDims(outShapes[0]);
return Ptr<BackendNode>(new InfEngineBackendNode(ieLayer));
}
#endif // HAVE_INF_ENGINE
private:
std::vector<MatShape> outShapes;
};
Ptr<ReshapeLayer> ReshapeLayer::create(const LayerParams& params)
{
return Ptr<ReshapeLayer>(new ReshapeLayerImpl(params));
}
}
}