root
/
MNN
mirror of https://github.com/alibaba/MNN.git
1
0
Fork 0
Code Issues Actions 7 Packages Projects Releases Wiki Activity
MNN/express/NeuralNetWorkOp.cpp

1450 lines
62 KiB
C++
Raw Blame History

This file contains ambiguous Unicode characters

This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.

//
// NeuralNetWorkOp.cpp
// MNN
//
// Created by MNN on 2019/06/27.
// Copyright © 2018, Alibaba Group Holding Limited
//
#include <algorithm>
#include <map>
#include <numeric>
#include <MNN/expr/ExprCreator.hpp>
#include <MNN/MNNDefine.h>
#include "MNN_generated.h"
#include "Utils.hpp"
namespace MNN {
namespace Express {
static PadMode _convertPadMode(PaddingMode mode) {
switch (mode) {
case CAFFE:
return PadMode_CAFFE;
case VALID:
return PadMode_VALID;
case SAME:
return PadMode_SAME;
default:
break;
}
return PadMode_CAFFE;
}
static PoolPadType _convertPoollingPadMode(PaddingMode mode) {
switch (mode) {
case CAFFE:
return PoolPadType_CAFFE;
case VALID:
return PoolPadType_VALID;
case SAME:
return PoolPadType_SAME;
default:
break;
}
return PoolPadType_CAFFE;
}
/*create a input variable.
Args:
shape: A vector, the shape of the variable.
data_format: A enum, NCHW/NHWC/NC4HW4 is allowed.
dtype: The type of the elements of the resulting variable.
Returns:
output: A variable.
*/
VARP _Input(INTS shape, Dimensionformat data_format, halide_type_t dtype) {
Variable::Info info;
info.dim = std::move(shape);
info.order = data_format;
info.type = dtype;
info.ptr = nullptr;
return (Variable::create(Expr::create(std::move(info))));
}
VARP _Scalar(const void* ptr, halide_type_t type) {
Variable::Info info;
info.dim = {};
info.order = NHWC;
info.type = type;
info.ptr = (void*)ptr;
return (Variable::create(Expr::create(std::move(info))));
}
/*create a constant variable.
Args:
ptr: A pointer. Indicates the values.
shape: A vector, the shape of the variable.
format: A enum, NCHW/NHWC/NC4HW4 is allowed.
type: The type of the elements of the resulting variable.
Returns:
output: A constant variable.
*/
VARP _Const(const void* ptr, INTS shape, Dimensionformat format, halide_type_t type) {
Variable::Info info;
info.dim = std::move(shape);
info.order = format;
info.type = type;
info.ptr = (void*)ptr;
return (Variable::create(Expr::create(std::move(info))));
}
VARP _Const(float value, INTS shape, Dimensionformat format) {
Variable::Info info;
info.dim = std::move(shape);
info.order = format;
info.type = halide_type_of<float>();
info.syncSize();
std::vector<float> values(info.size);
for (int i = 0; i < info.size; ++i) {
values[i] = value;
}
info.ptr = (void*)values.data();
return (Variable::create(Expr::create(std::move(info))));
}
VARP _TrainableParam(const void* ptr, INTS dims, Dimensionformat format, halide_type_t type) {
auto v = _Const(ptr, dims, format, type);
v.fix(VARP::TRAINABLE);
return v;
}
VARP _TrainableParam(float value, INTS dims, Dimensionformat format) {
auto v = _Const(value, dims, format);
v.fix(VARP::TRAINABLE);
return v;
}
VARP _Conv(VARP weight, VARP bias, VARP x, PaddingMode pad, INTS stride, INTS dilate, int group, INTS pads) {
std::unique_ptr<OpT> convOp(new OpT);
convOp->type = OpType_Convolution;
auto shape = weight->getInfo();
if (NHWC == shape->order) {
weight = _Transpose(weight, {0, 3, 1, 2});
shape = weight->getInfo();
}
auto channel = std::vector<int>{shape->dim[0], shape->dim[1]};
auto kernelSize = std::vector<int>{shape->dim[3], shape->dim[2]};
if (1 == channel[1] && channel[0] == group) {
convOp->type = OpType_ConvolutionDepthwise;
channel[1] = group;
}
convOp->main.type = OpParameter_Convolution2D;
convOp->main.value = new Convolution2DT;
auto conv2D = convOp->main.AsConvolution2D();
conv2D->common.reset(new Convolution2DCommonT);
if (pads.size() == 2) {
conv2D->common->padX = pads[0];
conv2D->common->padY = pads[1];
} else {
conv2D->common->pads = std::move(pads);
}
conv2D->common->padMode = _convertPadMode(pad);
conv2D->common->strideX = stride[0];
conv2D->common->strideY = stride[1];
conv2D->common->group = group;
conv2D->common->outputCount = channel[0];
conv2D->common->inputCount = channel[1];
conv2D->common->dilateX = dilate[0];
conv2D->common->dilateY = dilate[1];
conv2D->common->kernelX = kernelSize[0];
conv2D->common->kernelY = kernelSize[1];
if (nullptr == bias) {
return (Variable::create(Expr::create(convOp.get(), {x, weight})));
}
return (Variable::create(Expr::create(convOp.get(), {x, weight, bias})));
}
VARP _Conv(std::vector<float>&& weight, std::vector<float>&& bias, VARP x, INTS channel, INTS kernelSize,
PaddingMode pad, INTS stride, INTS dilate, int group, INTS pads, bool relu, bool relu6) {
std::unique_ptr<OpT> convOp(new OpT);
convOp->type = OpType_Convolution;
if (channel[0] == channel[1] && channel[0] == group) {
convOp->type = OpType_ConvolutionDepthwise;
}
convOp->main.type = OpParameter_Convolution2D;
convOp->main.value = new Convolution2DT;
auto conv2D = convOp->main.AsConvolution2D();
conv2D->common.reset(new Convolution2DCommonT);
conv2D->common->padMode = _convertPadMode(pad);
if (pads.size() == 2) {
conv2D->common->padX = pads[0];
conv2D->common->padY = pads[1];
} else {
conv2D->common->pads = std::move(pads);
}
conv2D->common->strideX = stride[0];
conv2D->common->strideY = stride[1];
conv2D->common->group = group;
conv2D->common->outputCount = channel[1];
conv2D->common->inputCount = channel[0];
conv2D->common->dilateX = dilate[0];
conv2D->common->dilateY = dilate[1];
conv2D->common->kernelX = kernelSize[0];
conv2D->common->kernelY = kernelSize[1];
conv2D->common->relu6 = relu6;
conv2D->common->relu = relu;
MNN_ASSERT(weight.size() == channel[1] * (channel[0] / group) * kernelSize[0] * kernelSize[1]);
conv2D->weight = std::move(weight);
MNN_ASSERT(bias.size() == channel[1]);
conv2D->bias = std::move(bias);
return (Variable::create(Expr::create(convOp.get(), {x})));
}
VARP _Conv(std::vector<int8_t>&& weight, std::vector<float>&& bias, VARP x, INTS channel, INTS kernelSize,
PaddingMode pad, INTS stride, INTS dilate, int group, INTS pads, bool relu, bool relu6) {
std::unique_ptr<OpT> convOp(new OpT);
convOp->type = OpType_Convolution;
if (channel[0] == channel[1] && channel[0] == group) {
convOp->type = OpType_ConvolutionDepthwise;
}
convOp->main.type = OpParameter_Convolution2D;
convOp->main.value = new Convolution2DT;
auto conv2D = convOp->main.AsConvolution2D();
conv2D->common.reset(new Convolution2DCommonT);
conv2D->common->padMode = _convertPadMode(pad);
if (pads.size() == 2) {
conv2D->common->padX = pads[0];
conv2D->common->padY = pads[1];
} else {
conv2D->common->pads = std::move(pads);
}
conv2D->common->strideX = stride[0];
conv2D->common->strideY = stride[1];
conv2D->common->group = group;
conv2D->common->outputCount = channel[1];
conv2D->common->inputCount = channel[0];
conv2D->common->dilateX = dilate[0];
conv2D->common->dilateY = dilate[1];
conv2D->common->kernelX = kernelSize[0];
conv2D->common->kernelY = kernelSize[1];
conv2D->common->relu6 = relu6;
conv2D->common->relu = relu;
MNN_ASSERT(weight.size() == channel[1] * (channel[0] / group) * kernelSize[0] * kernelSize[1]);
conv2D->quanParameter.reset(new IDSTQuanT);
conv2D->quanParameter->type = 3;
conv2D->quanParameter->buffer = std::move(weight);
conv2D->weight.clear();
MNN_ASSERT(bias.size() == channel[1]);
conv2D->bias = std::move(bias);
return (Variable::create(Expr::create(convOp.get(), {x})));
}
VARP _Conv(float weight, float bias, VARP x, INTS channel, INTS kernelSize, PaddingMode pad, INTS stride, INTS dilate,
int group) {
std::unique_ptr<OpT> convOp(new OpT);
convOp->type = OpType_Convolution;
if (channel[0] == channel[1] && channel[0] == group) {
convOp->type = OpType_ConvolutionDepthwise;
}
convOp->main.type = OpParameter_Convolution2D;
convOp->main.value = new Convolution2DT;
auto conv2D = convOp->main.AsConvolution2D();
conv2D->common.reset(new Convolution2DCommonT);
conv2D->common->padMode = _convertPadMode(pad);
conv2D->common->strideX = stride[0];
conv2D->common->strideY = stride[1];
conv2D->common->group = group;
conv2D->common->outputCount = channel[1];
conv2D->common->inputCount = channel[0];
conv2D->common->dilateX = dilate[0];
conv2D->common->dilateY = dilate[1];
conv2D->common->kernelX = kernelSize[0];
conv2D->common->kernelY = kernelSize[1];
conv2D->weight.resize(channel[1] * (channel[0] / group) * kernelSize[0] * kernelSize[1]);
std::fill(conv2D->weight.begin(), conv2D->weight.end(), weight);
conv2D->bias.resize(channel[1]);
std::fill(conv2D->bias.begin(), conv2D->bias.end(), bias);
return (Variable::create(Expr::create(convOp.get(), {x})));
}
VARP _Deconv(VARP weight, VARP bias, VARP x, PaddingMode pad, INTS stride, INTS dilate, int group, INTS pads) {
std::unique_ptr<OpT> convOp(new OpT);
convOp->type = OpType_Deconvolution;
auto shape = weight->getInfo();
auto channel = std::vector<int>{shape->dim[1], shape->dim[0]};
auto kernelSize = std::vector<int>{shape->dim[3], shape->dim[2]};
if (channel[1] * channel[0] == group) {
convOp->type = OpType_DeconvolutionDepthwise;
channel[1] = group;
channel[0] = group;
}
convOp->main.type = OpParameter_Convolution2D;
convOp->main.value = new Convolution2DT;
auto conv2D = convOp->main.AsConvolution2D();
conv2D->common.reset(new Convolution2DCommonT);
if (pads.size() == 2) {
conv2D->common->padX = pads[0];
conv2D->common->padY = pads[1];
} else {
conv2D->common->pads = std::move(pads);
}
conv2D->common->padMode = _convertPadMode(pad);
conv2D->common->strideX = stride[0];
conv2D->common->strideY = stride[1];
conv2D->common->group = group;
conv2D->common->outputCount = channel[0];
conv2D->common->inputCount = channel[1];
conv2D->common->dilateX = dilate[0];
conv2D->common->dilateY = dilate[1];
conv2D->common->kernelX = kernelSize[0];
conv2D->common->kernelY = kernelSize[1];
if (nullptr != bias) {
return (Variable::create(Expr::create(std::move(convOp), {x, weight, bias})));
}
return (Variable::create(Expr::create(std::move(convOp), {x, weight})));
}
static VARP _Pool(VARP x, INTS kernel, INTS stride, PoolType type, PaddingMode pad, INTS pads) {
std::unique_ptr<OpT> pool(new OpT);
pool->type = OpType_Pooling;
pool->main.type = OpParameter_Pool;
pool->main.value = new PoolT;
if (kernel[0] == -1 && kernel[1] == -1) {
pool->main.AsPool()->isGlobal = true;
}
pool->main.AsPool()->padX = 0;
pool->main.AsPool()->padY = 0;
if (pads.size() >= 2) {
pool->main.AsPool()->padX = pads[0];
pool->main.AsPool()->padY = pads[1];
}
pool->main.AsPool()->padType = _convertPoollingPadMode(pad);
pool->main.AsPool()->kernelX = kernel[0];
pool->main.AsPool()->kernelY = kernel[1];
pool->main.AsPool()->strideX = stride[0];
pool->main.AsPool()->strideY = stride[1];
pool->main.AsPool()->type = type;
return (Variable::create(Expr::create(pool.get(), {x})));
}
VARP _AvePool(VARP x, INTS kernel, INTS stride, PaddingMode pad, INTS pads) {
return _Pool(x, kernel, stride, PoolType_AVEPOOL, pad, pads);
}
VARP _MaxPool(VARP x, INTS kernel, INTS stride, PaddingMode pad, INTS pads) {
return _Pool(x, kernel, stride, PoolType_MAXPOOL, pad, pads);
}
/*Reshapes a variable.
Args:
x: A variable.
shape: A vector, the shape of the target variable.
original_format: A enum, only NCHW/NHWC is allowed, NC4HW4 is not allowed,
as it provides additional information(x comes from NCHW or NHWC) When x is NC4HW4.
Returns:
output: A variable with the same type as `x`.
*/
VARP _Reshape(VARP x, INTS shape, Dimensionformat original_format) {
std::unique_ptr<OpT> reshape(new OpT);
reshape->type = OpType_Reshape;
reshape->main.type = OpParameter_Reshape;
reshape->main.value = new ReshapeT;
reshape->main.AsReshape()->dims = shape;
reshape->main.AsReshape()->dimType = (MNN_DATA_FORMAT)Utils::convertFormat(original_format);
return (Variable::create(Expr::create(reshape.get(), {x})));
}
/*Reshapes a variable.
Args:
x: A variable.
shape: A variable, the shape of the target variable.
Returns:
output: A variable with the same type as `x`.
*/
VARP _Reshape(VARP x, VARP shape) {
MNN_ASSERT(nullptr != x);
MNN_ASSERT(nullptr != x->getInfo());
std::unique_ptr<OpT> reshape(new OpT);
reshape->type = OpType_Reshape;
reshape->main.type = OpParameter_Reshape;
reshape->main.value = new ReshapeT;
reshape->main.AsReshape()->dimType = (MNN_DATA_FORMAT)Utils::convertFormat(x->getInfo()->order);
return (Variable::create(Expr::create(reshape.get(), {x, shape})));
}
VARP _Scale(VARP x, int channels, std::vector<float>&& scales, std::vector<float>&& bias) {
std::unique_ptr<OpT> scale(new OpT);
scale->type = OpType_Scale;
scale->main.type = OpParameter_Scale;
scale->main.value = new ScaleT;
scale->main.AsScale()->channels = channels;
scale->main.AsScale()->scaleData = std::move(scales);
scale->main.AsScale()->biasData = std::move(bias);
return (Variable::create(Expr::create(std::move(scale), {x})));
}
/*Given an input value x, it computes the output as x if x > 0 and slope * x if x <= 0.
Args:
x: A variable.
slope: A float, a positive float value, it leakes the negative part by multiplying with `slope` rather than setting it to 0.0f.
Returns:
output: A variable with the same type as `x`.
*/
VARP _Relu(VARP x, float slope) {
std::unique_ptr<OpT> relu(new OpT);
relu->type = OpType_ReLU;
relu->main.type = OpParameter_Relu;
relu->main.value = new ReluT;
relu->main.AsRelu()->slope = slope;
return (Variable::create(Expr::create(relu.get(), {x})));
}
/*Given an input value x， it computes Rectified Linear 6: min(max(x, 0), 6).
Args:
x: A variable.
Returns:
output: A variable with the same type as `x`.
*/
VARP _Relu6(VARP x) {
std::unique_ptr<OpT> relu(new OpT);
relu->type = OpType_ReLU6;
return (Variable::create(Expr::create(relu.get(), {x})));
}
/*Given an input value x, it computes the output as x if x > 0 and slopes * x if x <= 0.
Args:
x: A variable, must be 4-D with NC4HW4 format.
slopes: A vector, has save size as x.
Returns:
output: A variable with the same type as `x`.
*/
VARP _PRelu(VARP x, std::vector<float>&& slopes) {
std::unique_ptr<OpT> prelu(new OpT);
prelu->type = OpType_PReLU;
prelu->main.type = OpParameter_PRelu;
prelu->main.value = new PReluT;
prelu->main.AsPRelu()->slope = slopes;
prelu->main.AsPRelu()->slopeCount = slopes.size();
return (Variable::create(Expr::create(prelu.get(), {x})));
}
/*Computes softmax activations.
Args:
logits: A non-empty variable. Must be Halide_Type_Float.
axis: The dimension softmax would be performed on. The default is -1 which indicates the last dimension.
Returns:
output: A variable with the same type as `logits`.
*/
VARP _Softmax(VARP logits, int axis) {
std::unique_ptr<OpT> softmax(new OpT);
softmax->type = OpType_Softmax;
softmax->main.type = OpParameter_Axis;
softmax->main.value = new AxisT;
softmax->main.AsAxis()->axis = axis;
return (Variable::create(Expr::create(softmax.get(), {logits})));
}
/*Computes softplus: log(exp(features) + 1).
Args:
features: A variable. Must be Halide_Type_Float.
Returns:
A variable with the same type as `features`.
*/
VARP _Softplus(VARP features) {
return _Log(_Add(_Exp(features), _Const(1)));
}
/*Computes softsign: features / (abs(features) + 1).
Args:
features: A variable. Must be Halide_Type_Float.
Returns:
A variable with the same type as `features`.
*/
VARP _Softsign(VARP features) {
return _Divide(features, _Add(_Abs(features), _Const(1)));
}
/*Concatenates variables along one dimension.
Args:
values: A list of variables a single variable.
axis: A int. Dimension along which to concatenate.
Must be in the range [-rank(values), rank(values)).
As in Python, indexing for axis is 0-based.
Positive axis in the rage of [0, rank(values)) refers to axis-th dimension.
And negative axis refers to axis + rank(values)-th dimension.
Returns:
A variable resulting from concatenation of the input variables.
*/
VARP _Concat(VARPS values, int axis) {
std::unique_ptr<OpT> concat(new OpT);
concat->type = OpType_Concat;
concat->main.type = OpParameter_Axis;
concat->main.value = new AxisT;
concat->main.AsAxis()->axis = axis;
return (Variable::create(Expr::create(concat.get(), values)));
}
/*Convert a variable to another format(possibily added after `input`).
Args:
input: A variable.
format: The target format.
Returns:
A variable. If `input` is already `format`, then return `input` directly, otherwize add a variable after `input` with `format`.
*/
VARP _Convert(VARP input, Dimensionformat format) {
if (nullptr != input->getInfo()) {
auto source = input->getInfo()->order;
if (source == format) {
return input;
}
}
std::unique_ptr<OpT> convert(new OpT);
convert->type = OpType_ConvertTensor;
convert->main.type = OpParameter_TensorConvertInfo;
convert->main.value = new TensorConvertInfoT;
convert->main.AsTensorConvertInfo()->dest = (MNN_DATA_FORMAT)Utils::convertFormat(format);
return (Variable::create(Expr::create(convert.get(), {input})));
}
/*Splits a variable value into a list of sub variables.
Args:
value: The variable to split.
size_splits: A vector, a 1-D integer containing the sizes of each output variable along axis.
axis: A int, the dimension along which to split. Must be in the range [-rank(value), rank(value)). Defaults to 0
Returns:
A list of variables.
*/
std::vector<VARP> _Split(VARP value, INTS size_splits, int axis) {
MNN_ASSERT(size_splits.size() >= 1);
std::unique_ptr<OpT> op(new OpT);
op->type = OpType_Slice;
op->main.type = OpParameter_Slice;
op->main.value = new SliceT;
op->main.AsSlice()->axis = axis;
op->main.AsSlice()->sourceType = NetSource_TENSORFLOW;
op->main.AsSlice()->slicePoints = size_splits;
int slices = size_splits.size() == 1 ? size_splits[0] : (int)size_splits.size();
EXPRP expr = Expr::create(std::move(op), {value}, slices);
std::vector<VARP> res;
for (int i = 0; i < slices; ++i) {
res.emplace_back(Variable::create(expr, i));
}
return res;
}
VARP _Slice(VARP x, VARP starts, VARP sizes) {
std::unique_ptr<OpT> slice(new OpT);
slice->type = OpType_SliceTf;
return (Variable::create(Expr::create(slice.get(), {x, starts, sizes})));
}
VARP _StridedSlice(VARP input, VARP begin, VARP end, VARP strided, int32_t beginMask,
int32_t endMask, int32_t ellipsisMask, int32_t newAxisMask, int32_t shrinkAxisMask) {
std::unique_ptr<OpT> op(new OpT);
op->type = OpType_StridedSlice;
op->main.type = OpParameter_StridedSliceParam;
op->main.value = new StridedSliceParamT;
op->main.AsStridedSliceParam()->T = DataType_DT_FLOAT;
op->main.AsStridedSliceParam()->beginMask = beginMask;
op->main.AsStridedSliceParam()->endMask = endMask;
op->main.AsStridedSliceParam()->ellipsisMask = ellipsisMask;
op->main.AsStridedSliceParam()->newAxisMask = newAxisMask;
op->main.AsStridedSliceParam()->shrinkAxisMask = shrinkAxisMask;
return (Variable::create(Expr::create(op.get(), {input, begin, end, strided})));
}
/*Transposes x.
Args:
x: A variable.
perm: A vector, indicating the permutation of the dimensions of x.
Returns:
A transposed variable.
*/
VARP _Transpose(VARP x, INTS perm) {
auto permVar = _Const((const void*)perm.data(), {static_cast<int>(perm.size())}, NHWC, halide_type_of<int>());
return _Transpose(x, permVar);
}
VARP _Transpose(VARP x, VARP perm) {
std::unique_ptr<OpT> transpose(new OpT);
transpose->type = OpType_Transpose;
transpose->main.type = OpParameter_Transpose;
transpose->main.value = new TransposeT;
transpose->main.AsTranspose()->Tperm = DataType_DT_INT32;
return (Variable::create(Expr::create(std::move(transpose), {x, perm})));
}
VARP _ChannelShuffle(VARP x, int group) {
x = _Convert(x, NHWC);
x = _Reshape(x, {0, 0, 0, group, -1}, NHWC);
x = _Transpose(x, {0, 1, 2, 4, 3});
x = _Reshape(x, {0, 0, 0, -1}, NHWC);
x = _Convert(x, NC4HW4);
return x;
}
VARP _ReverseSequence(VARP x, VARP y, int batchDim, int seqDim) {
std::unique_ptr<OpT> op(new OpT);
op->type = OpType_ReverseSequence;
op->main.type = OpParameter_ReverseSequenceParam;
op->main.value = new ReverseSequenceParamT;
op->main.AsReverseSequenceParam()->batchDim = batchDim;
op->main.AsReverseSequenceParam()->seqDim = seqDim;
return (Variable::create(Expr::create(op.get(), {x, y})));
}
/*Convert a variable to another format(possibily added before `input`).
Args:
input: A variable.
format: The target format.
Returns:
A variable. If `input` is already `format`, then return `input` directly, otherwize add a variable before `input` with `format`.
*/
VARP _ChangeInputFormat(VARP input, Dimensionformat format) {
if (nullptr == input || nullptr == input->getInfo()) {
return nullptr;
}
if (input->getInfo()->order == format) {
return input;
}
auto input_before = _Input(input->getInfo()->dim, format, input->getInfo()->type);
auto convert = _Convert(input_before, input->getInfo()->order);
Variable::replace(input, convert);
return input_before;
}
VARP _Clone(VARP source, bool deepCopy) {
if (nullptr == source || nullptr == source->expr().first) {
return nullptr;
}
if (!deepCopy) {
return Variable::create(source->expr().first, source->expr().second);
}
auto info = source->getInfo();
auto sourcePtr = source->readMap<void>();
if (nullptr == info || nullptr == sourcePtr) {
MNN_ERROR("Source Buffer Not Available\n");
return nullptr;
}
auto inputVar = _Input(info->dim, info->order, info->type);
auto destPtr = inputVar->writeMap<void>();
if (nullptr == destPtr) {
MNN_ERROR("Alloc Buffer Error\n");
return nullptr;
}
::memcpy(destPtr, sourcePtr, info->size * info->type.bytes());
return inputVar;
}
VARP _Conv2DBackPropFilter(VARP input, VARP inputGrad, INTS kernelSize, PaddingMode pad, INTS stride, INTS dilate,
int group, INTS pads) {
std::unique_ptr<OpT> convOp(new OpT);
convOp->type = OpType_Conv2DBackPropFilter;
auto srcShape = input->getInfo();
auto dstShape = inputGrad->getInfo();
auto channel = std::vector<int>{srcShape->dim[1], dstShape->dim[1]};
convOp->main.type = OpParameter_Convolution2D;
convOp->main.value = new Convolution2DT;
auto conv2D = convOp->main.AsConvolution2D();
conv2D->common.reset(new Convolution2DCommonT);
conv2D->common->padX = pads[0];
conv2D->common->padY = pads[1];
conv2D->common->padMode = _convertPadMode(pad);
conv2D->common->strideX = stride[0];
conv2D->common->strideY = stride[1];
conv2D->common->group = group;
conv2D->common->outputCount = channel[1];
conv2D->common->inputCount = channel[0];
conv2D->common->dilateX = dilate[0];
conv2D->common->dilateY = dilate[1];
conv2D->common->kernelX = kernelSize[0];
conv2D->common->kernelY = kernelSize[1];
INTS weightDims = {channel[1], channel[0] / group, kernelSize[1], kernelSize[0]};
return Variable::create(Expr::create(std::move(convOp), {input, inputGrad}));
}
VARP _PoolGrad(VARP originInput, VARP originOutput, VARP inputGrad, INTS kernel, INTS stride, PoolingMode type,
PaddingMode pad, INTS pads) {
std::unique_ptr<OpT> pool(new OpT);
pool->type = OpType_PoolGrad;
pool->main.type = OpParameter_Pool;
pool->main.value = new PoolT;
if (kernel[0] == -1 && kernel[1] == -1) {
pool->main.AsPool()->isGlobal = true;
}
pool->main.AsPool()->padX = 0;
pool->main.AsPool()->padY = 0;
if (pads.size() >= 2) {
pool->main.AsPool()->padX = pads[0];
pool->main.AsPool()->padY = pads[1];
}
pool->main.AsPool()->padType = _convertPoollingPadMode(pad);
pool->main.AsPool()->kernelX = kernel[0];
pool->main.AsPool()->kernelY = kernel[1];
pool->main.AsPool()->strideX = stride[0];
pool->main.AsPool()->strideY = stride[1];
pool->main.AsPool()->type = (PoolType)type;
return (Variable::create(Expr::create(std::move(pool), {originInput, originOutput, inputGrad})));
}
/*Crop images.
Args:
images: 4-D variable of NC4HW4 format.
size: A variable. It takes the shape of `size` as output cropped variable's shape while omits the values/format of `size`.
axis: A int indicating the dimention to crop. Must be >=2. All dimensions up to but excluding `axis` are preserved, while the dimensions including and trailing `axis` are cropped.
offset: A vector of int indicating the offsets. length(`offset`) must be >=1 and <=2. If length(`offset`) is 1, then all dimensions are offset by this amount.Otherwise, the number of offsets must equal the number of cropped axes in each dimension accordingly.
Returns:
The cropped 4-D variable of NC4HW4 format.
*/
VARP _Crop(VARP images, VARP size, int axis, INTS offset) {
std::unique_ptr<OpT> crop(new OpT);
crop->type = OpType_Crop;
crop->main.type = OpParameter_Crop;
crop->main.value = new CropT;
crop->main.AsCrop()->axis = axis;
crop->main.AsCrop()->offset = offset;
return (Variable::create(Expr::create(std::move(crop), {images, size})));
}
/*Resize images.
Args:
images: 4-D variable of NC4HW4 format.
xScale: A float.
yScale: A float.
Returns:
The resized 4-D variable of NC4HW4 format.
*/
VARP _Resize(VARP images, float xScale, float yScale) {
std::unique_ptr<OpT> resize(new OpT);
resize->type = OpType_Resize;
resize->main.type = OpParameter_Resize;
resize->main.value = new ResizeT;
resize->main.AsResize()->xScale = xScale;
resize->main.AsResize()->yScale = yScale;
return (Variable::create(Expr::create(std::move(resize), {images})));
}
/*Pads a variable.
Args:
x: A variable.
paddings: A variable of type Halide_Type_Int. The shape is [n, 2] where n is the rank of variable.
mode: A enum, One of PadValueMode_CONSTANT, PadValueMode_SYMMETRIC, or PadValueMode_REFLECT.
Returns:
A variable. Has the same type as x.
*/
VARP _Pad(VARP x, VARP paddings, PadValueMode mode) {
std::unique_ptr<OpT> pad(new OpT);
pad->type = OpType_Padding;
pad->main.type = OpParameter_PadParam;
pad->main.value = new PadParamT;
switch (mode) {
case CONSTANT:
pad->main.AsPadParam()->mode = MNN::PadValueMode_CONSTANT;
break;
case SYMMETRIC:
pad->main.AsPadParam()->mode = MNN::PadValueMode_SYMMETRIC;
break;
case REFLECT:
pad->main.AsPadParam()->mode = MNN::PadValueMode_REFLECT;
break;
default:
pad->main.AsPadParam()->mode = MNN::PadValueMode_CONSTANT;
break;
}
return (Variable::create(Expr::create(std::move(pad), {x, paddings})));
}
/*Returns a variable with an additional dimension inserted at index axis.
Args:
input: A variable.
axis: A int, specifying the dimension index at which to expand the shape of input.
Given an input of D dimensions, axis must be in range [-(D+1), D] (inclusive).
Returns:
A variable with the same data as input, with an additional dimension inserted at the index specified by axis.
*/
VARP _ExpandDims(VARP input, int axis) {
std::unique_ptr<OpT> expand(new OpT);
expand->type = OpType_ExpandDims;
expand->main.type = OpParameter_ExpandDims;
expand->main.value = new ExpandDimsT;
expand->main.AsExpandDims()->axis = axis;
return (Variable::create(Expr::create(std::move(expand), {input})));
}
VARP _ExpandDims(VARP input, VARP axis) {
std::unique_ptr<OpT> expand(new OpT);
expand->type = OpType_ExpandDims;
expand->main.type = OpParameter_ExpandDims;
expand->main.value = new ExpandDimsT;
return (Variable::create(Expr::create(std::move(expand), {input, axis})));
}
/*Returns the shape of a variable.
Args:
input: A variable.
Returns:
A variable of Halide_Type_Int.
*/
VARP _Shape(VARP input) {
std::unique_ptr<OpT> shape(new OpT);
shape->type = OpType_Shape;
return (Variable::create(Expr::create(std::move(shape), {input})));
}
/*Stacks a list of rank-R variables into one rank-(R+1) variable.
Packs the list of variables in `values` into a ariable with rank one higher than each variable in values,
by packing them along the axis dimension.
Given a list of length N of variables of shape (A, B, C);
if axis == 0 then the output variable will have the shape (N, A, B, C).
if axis == 1 then the output variable will have the shape (A, N, B, C). Etc.
Args:
values: A list of variable objects with the same shape and type.
axis: An int. The axis to stack along. Defaults to the first dimension. Negative values wrap around,
so the valid range is [-(R+1), R+1).
Returns:
output: A stacked variable with the same type as `values`.
*/
VARP _Stack(VARPS values, int axis) {
std::unique_ptr<OpT> pack(new OpT);
pack->type = OpType_Pack;
pack->main.type = OpParameter_PackParam;
pack->main.value = new PackParamT;
pack->main.AsPackParam()->axis = axis;
return (Variable::create(Expr::create(std::move(pack), values)));
}
/*Extracts crops from the input image variable and resizes them using bilinear sampling or nearest neighbor sampling (possibly with aspect ratio change)
to a common output size specified by crop_size.
Returns a variable with crops from the input image at positions defined at the bounding box locations in boxes.
The cropped boxes are all resized (with bilinear or nearest neighbor interpolation) to a fixed size = [crop_height, crop_width].
The result is a 4-D tensor [num_boxes, crop_height, crop_width, depth](supposing NHWC format).
Arguments:
image: A 4-D variable of shape [batch, image_height, image_width, depth](supposing NHWC format). Both image_height and image_width need to be positive.
boxes: A 2-D variable of shape [num_boxes, 4]. The i-th row of the variable specifies the coordinates of a box in the box_ind[i] image and is specified in normalized coordinates [y1, x1, y2, x2].
A normalized coordinate value of y is mapped to the image coordinate at y * (image_height - 1), so as the [0, 1] interval of normalized image height is mapped to [0, image_height - 1] in image height coordinates. We do allow y1 > y2, in which case the sampled crop is an up-down flipped version of the original image. The width dimension is treated similarly. Normalized coordinates outside the [0, 1] range are allowed, in which case we use extrapolation_value to extrapolate the input image values.
box_ind: A 1-D variable of shape [num_boxes] with int values in [0, batch). The value of box_ind[i] specifies the image that the i-th box refers to.
crop_size: A 1-D variable of 2 elements, size = [crop_height, crop_width]. All cropped image patches are resized to this size. The aspect ratio of the image content is not preserved. Both crop_height and crop_width need to be positive.
method: A enum, either CropAndResizeMethod_NEAREST, or CropAndResizeMethod_BILINEAR, default to CropAndResizeMethod_BILINEAR.
extrapolation_value: Value used for extrapolation, when applicable.
Returns:
Output: A 4-D variable of shape [num_boxes, crop_height, crop_width, depth](supposing NHWC format).
*/
VARP _CropAndResize(VARP image, VARP boxes, VARP box_ind, VARP crop_size, InterpolationMethod method, float extrapolation_value) {
std::unique_ptr<OpT> car(new OpT);
car->type = OpType_CropAndResize;
car->main.type = OpParameter_CropAndResize;
car->main.value = new CropAndResizeT;
car->main.AsCropAndResize()->extrapolationValue = extrapolation_value;
switch (method) {
case NEAREST:
car->main.AsCropAndResize()->method = CropAndResizeMethod_NEAREST;
break;
case BILINEAR:
default:
car->main.AsCropAndResize()->method = CropAndResizeMethod_BILINEAR;
break;
}
return (Variable::create(Expr::create(std::move(car), {image, boxes, box_ind, crop_size})));
}
/*Creates a variable filled with a scalar value.
Args:
dims: A variable. Must be 1-D Halide_Type_Int. Represents the shape of the output variable.
value: A variable. 0-D (scalar). Value to fill the returned variable.
Returns:
A variable. Has the same type as value.
*/
VARP _Fill(VARP dims, VARP value) {
std::unique_ptr<OpT> fill(new OpT);
fill->type = OpType_Fill;
fill->main.type = OpParameter_Fill;
fill->main.value = new FillT;
return (Variable::create(Expr::create(std::move(fill), {dims, value})));
}
/*Constructs a variable by tiling a given variable.
Args:
input: A variable. 1-D or higher.
multiples: A variable. Must be 1-D Halide_Type_Int.Length must be the same as the number of dimensions in input.
Returns:
A variable. Has the same type as input.
*/
VARP _Tile(VARP input, VARP multiples) {
std::unique_ptr<OpT> tile(new OpT);
tile->type = OpType_Tile;
return (Variable::create(Expr::create(std::move(tile), {input, multiples})));
}
/*Gather slices from params according to indices.
Arguments:
params: The variable from which to gather values.
indices: Index variable. Must be Halide_Type_Int in range [0, ndims(params)-1].
Returns:
Output: Values from params gathered from indices given by indices.
*/
VARP _Gather(VARP params, VARP indices) {
std::unique_ptr<OpT> gather(new OpT);
gather->type = OpType_Gather;
gather->main.value = new GatherT;
return (Variable::create(Expr::create(std::move(gather), {params, indices})));
}
/*Gather slices from params axis according to indices.
Arguments:
params: The variable from which to gather values.
indices: Index variable. Must be Halide_Type_Int in range [0, ndims(params)-1].
axis: A int, the axis in params to gather indices from. Supports negative indexes.
If set to 0, it's same as _Gather. Currently only 0 is supported.
Returns:
Output: Values from params gathered from indices given by indices.
*/
VARP _GatherV2(VARP params, VARP indices, VARP axis) {
std::unique_ptr<OpT> gather(new OpT);
gather->type = OpType_GatherV2;
gather->main.type = OpParameter_GatherV2;
gather->main.value = new GatherV2T;
if (axis.get()) {
return (Variable::create(Expr::create(std::move(gather), {params, indices, axis})));
} else {
return (Variable::create(Expr::create(std::move(gather), {params, indices})));
}
}
/*Removes dimensions of size 1 from the shape of a variable.
Args:
input: A variable. The input to squeeze.
axis: A vector, Defaults to {}. If specified, only squeezes the dimensions listed. The dimension index starts at 0.
Must be in the range [-rank(input), rank(input)).
Returns:
A variable. Has the same type as input. Contains the same data as input, but has one or more dimensions of size 1 removed.
*/
VARP _Squeeze(VARP input, INTS axis) {
std::unique_ptr<OpT> squeeze(new OpT);
squeeze->type = OpType_Squeeze;
auto squeezeParam = new SqueezeParamT;
squeezeParam->squeezeDims = axis;
squeeze->main.type = OpParameter_SqueezeParam;
squeeze->main.value = squeezeParam;
return Variable::create(Expr::create(std::move(squeeze), {input}));
}
VARP _Unsqueeze(VARP input, INTS axis) {
std::unique_ptr<OpT> unsqueeze(new OpT);
unsqueeze->type = OpType_Unsqueeze;
auto squeezeParam = new SqueezeParamT;
squeezeParam->squeezeDims = axis;
unsqueeze->main.type = OpParameter_SqueezeParam;
unsqueeze->main.value = squeezeParam;
return Variable::create(Expr::create(std::move(unsqueeze), {input}));
}
/*Computes exponential linear: alpha * (exp(features) - 1) if < 0, features otherwise.
features: A variable of type Halide_Type_Float
alpha: Alpha factor (positive float)
Returns:
A variable. Has the same type as features.
*/
VARP _Elu(VARP features, float alpha) {
std::unique_ptr<OpT> op(new OpT);
op->type = OpType_ELU;
auto eluParam = new ELUT;
op->main.type = OpParameter_ELU;
eluParam->alpha = alpha;
op->main.value = eluParam;
return (Variable::create(Expr::create(std::move(op), {features})));
}
/*Computes the size of the variable
Args:
input: A variable of type Halide_Type_Float or Halide_Type_Int
Returns:
A variable. The shape is (), and type is Halide_Type_Int
*/
VARP _Size(VARP input) {
std::unique_ptr<OpT> op(new OpT);
op->type = OpType_Size;
return (Variable::create(Expr::create(std::move(op), {input})));
}
/*Computes scaled exponential linear: scale * alpha * (exp(features) - 1) if < 0, scale * features otherwise.
Args:
features: A variable of type Halide_Type_Float
scale: Scaling factor (positive float)
alpha: Alpha factor (positive float)
Returns:
A variable. Has the same type as features.
*/
VARP _Selu(VARP features, float scale, float alpha) {
std::unique_ptr<OpT> op(new OpT);
op->type = OpType_Selu;
auto seluParam = new SeluT;
op->main.type = OpParameter_Selu;
seluParam->scale = scale;
seluParam->alpha = alpha;
op->main.value = seluParam;
return (Variable::create(Expr::create(std::move(op), {features})));
}
/*Gather slices from params into a variable with shape specified by indices.
Args:
params: A variable. The variables from which to gather values.
indices: A variable. Must be one of the following types: Halide_Type_Int.
Returns:
A variable. Has the same type as params.
*/
VARP _GatherND(VARP params, VARP indices) {
std::unique_ptr<OpT> op(new OpT);
op->type = OpType_GatherND;
return (Variable::create(Expr::create(std::move(op), {params, indices})));
}
/*BatchToSpace for N-D variables
This operation reshapes the "batch" dimension 0 into M + 1 dimensions of shape block_shape + [batch],
interleaves these blocks back into the grid defined by the spatial dimensions [1, ..., M],
to obtain a result with the same rank as the input.
The spatial dimensions of this intermediate result are then optionally cropped according to crops to
produce the output. This is the reverse of SpaceToBatch. See below for a precise description.
Arguments:
input: must be 4-D with NC4HW4 format. N-D with shape input_shape = [batch] + spatial_shape + remaining_shape, where spatial_shape has M dimensions.
block_shape: 1-D with shape [M], all values must be >= 1.
crops: 2-D with shape [M, 2], all values must be >= 0. crops[i] = [crop_start, crop_end] specifies the amount to crop from input dimension i + 1,
which corresponds to spatial dimension i. It is required that crop_start[i] + crop_end[i] <= block_shape[i] * input_shape[i + 1].
This operation is equivalent to the following steps:
Reshape input to reshaped of shape: [block_shape[0], ..., block_shape[M-1], batch / prod(block_shape),
input_shape[1], ..., input_shape[N-1]]
Permute dimensions of reshaped to produce permuted of shape
[batch / prod(block_shape),input_shape[1], block_shape[0], ..., input_shape[M], block_shape[M-1],input_shape[M+1], ..., input_shape[N-1]]
Reshape permuted to produce reshaped_permuted of shape
[batch / prod(block_shape),input_shape[1] * block_shape[0], ..., input_shape[M] * block_shape[M-1],input_shape[M+1], ..., input_shape[N-1]]
Crop the start and end of dimensions [1, ..., M] of reshaped_permuted according to crops to produce the output of shape:
[batch / prod(block_shape),input_shape[1] * block_shape[0] - crops[0,0] - crops[0,1], ..., input_shape[M] * block_shape[M-1] - crops[M-1,0] - crops[M-1,1],input_shape[M+1], ..., input_shape[N-1]]
Some examples:
for the following input of shape [4, 1, 1, 3], block_shape = [2, 2], and crops = [[0, 0], [0, 0]]:
[[[[1, 2, 3]]], [[[4, 5, 6]]], [[[7, 8, 9]]], [[[10, 11, 12]]]]
The output variable has shape [1, 2, 2, 3] and value:
x = [[[[1, 2, 3], [4, 5, 6]],
[[7, 8, 9], [10, 11, 12]]]]
Returns:
Output: The output variable
*/
VARP _BatchToSpaceND(VARP input, VARP block_shape, VARP crops) {
std::unique_ptr<OpT> op(new OpT);
std::unique_ptr<BlobT> blob_blockShape(new BlobT);
std::unique_ptr<BlobT> blob_paddings(new BlobT);
auto info_block_shape = block_shape->getInfo();
auto info_crops = crops->getInfo();
MNN_ASSERT(info_block_shape != nullptr);
MNN_ASSERT(info_crops != nullptr);
MNN_ASSERT(halide_type_int == info_block_shape->type.code);
MNN_ASSERT(halide_type_int == info_crops->type.code);
blob_blockShape->dims = info_block_shape->dim;
blob_blockShape->dataFormat = (MNN_DATA_FORMAT)Utils::convertFormat(info_block_shape->order);
blob_blockShape->dataType = (MNN::DataType)Utils::convertDataType(info_block_shape->type);
auto data_block_shape = block_shape->readMap<int>();
for (int i=0; i<info_block_shape->size; i++)
{
blob_blockShape->int32s.emplace_back(data_block_shape[i]);
}
blob_paddings->dims = info_crops->dim;
blob_paddings->dataFormat = (MNN_DATA_FORMAT)Utils::convertFormat(info_crops->order);
blob_paddings->dataType = (MNN::DataType)Utils::convertDataType(info_crops->type);
auto data_crop = crops->readMap<int>();
for (int i=0; i<info_crops->size; i++)
{
blob_paddings->int32s.emplace_back(data_crop[i]);
}
op->main.type = OpParameter_SpaceBatch;
op->type = OpType_BatchToSpaceND;
op->main.value = new SpaceBatchT;
op->main.AsSpaceBatch()->blockShape = std::move(blob_blockShape);
op->main.AsSpaceBatch()->padding = std::move(blob_paddings);
return Variable::create(Expr::create(std::move(op), {input}));
}
/*Copies a variable setting everything outside a central band in each innermost matrix.
Arguments:
input: Rank k variable.
num_lower: Number of subdiagonals to keep. If negative, keep entire lower triangle.
num_upper: Number of superdiagonals to keep. If negative, keep entire upper triangle.
Returns:
Output: Rank k variable of the same shape as input. The extracted banded tensor.
*/
VARP _MatrixBandPart(VARP input, VARP num_lower, VARP num_upper) {
std::unique_ptr<OpT> op(new OpT);
op->type = OpType_MatrixBandPart;
op->main.type = OpParameter_NONE;
return (Variable::create(Expr::create(std::move(op), {input, num_lower, num_upper})));
}
/*Calculates the mean and variance of x.
Args:
x: A variable. must be 4-D with NC4HW4 format.
axes: Array of ints. Axes along which to compute mean and variance. Ignored for this implementation: must be {2, 3}
shift: Not used in the current implementation.
keepdims: produce moments with the same dimensionality as the input. Ignored for this implementation: must be true.
Returns:
Two variable objects: mean and variance.
*/
std::vector<VARP> _Moments(VARP x, INTS axis, VARP shift, bool keepDims) {
std::unique_ptr<OpT> op(new OpT);
axis = {2, 3};
keepDims = true;
// if axis != {2,3} or keepDims != true, print warning.
// ignore shift.
op->type = OpType_Moments;
auto momentsParam = new MomentsParamT;
op->main.type = OpParameter_MomentsParam;
momentsParam->dim = axis;
momentsParam->keepDims = keepDims;
momentsParam->dType = (MNN::DataType)Utils::convertDataType(x->getInfo()->type);
op->main.value = momentsParam;
EXPRP expr = Expr::create(std::move(op), {x}, 2);
std::vector<VARP> res;
res.emplace_back(Variable::create(expr, 0));
res.emplace_back(Variable::create(expr, 1));
return res;
}
/*Computes the difference between two lists of numbers or strings.
Given a list x and a list y, this operation returns a list out that represents all values that are in x but not in y.
The returned list out is sorted in the same order that the numbers appear in x (duplicates are preserved).
This operation also returns a list idx that represents the position of each out element in x.
Arguments:
x: 1-D variable of type Halide_Type_Int. Values to keep.
y: 1-D variable of type Halide_Type_Int. Values to remove.
Returns:
Output out: 1-D variable of type Halide_Type_Int. Values present in x but not in y.
*/
VARP _SetDiff1D(VARP x, VARP y) {
std::unique_ptr<OpT> op(new OpT);
op->type = OpType_SetDiff1D;
op->main.type = OpParameter_NONE;
op->main.value = nullptr;
return Variable::create(Expr::create(std::move(op), {x, y}));
}
/*Rearranges blocks of spatial data, into depth.
More specifically, it outputs a copy of the input variable where values from the height and width dimensions are moved to the depth dimension.
The block_size indicates the input block size.
Non-overlapping blocks of size block_size x block_size are rearranged into depth at each location.
The depth of the output variable is block_size * block_size * input_depth.
The Y, X coordinates within each block of the input become the high order component of the output channel index.
The input variable's height and width must be divisible by block_size
Args:
input: A variable.
block_size: An int that is >= 2. The size of the spatial block.
Returns:
A variable. Has the same type as input.
*/
VARP _SpaceToDepth(VARP input, int block_size) {
std::unique_ptr<OpT> op(new OpT);
op->type = OpType_SpaceToDepth;
auto param = new DepthSpaceParamT;
param->blockSize = block_size;
op->main.type = OpParameter_DepthSpaceParam;
op->main.value = param;
return Variable::create(Expr::create(std::move(op), {input}));
}
/*This operation divides "spatial" dimensions [1, ..., M] of the input into a grid of blocks of shape block_shape,
and interleaves these blocks with the "batch" dimension
such that in the output, the spatial dimensions [1, ..., M] correspond to the position within the grid,
and the batch dimension combines both the position within a spatial block and the original batch position.
Prior to division into blocks, the spatial dimensions of the input are optionally zero padded according to paddings.
See below for a precise description.
Args:
input: A variable. must be 4-D with NC4HW4 format. N-D with shape input_shape = [batch] + spatial_shape + remaining_shape, where spatial_shape has M dimensions.
block_shape: A variable. Must be one of the following types: int32, int64. 1-D with shape [M], all values must be >= 1.
paddings: A variable. Must be one of the following types: int32, int64. 2-D with shape [M, 2], all values must be >= 0. paddings[i] = [pad_start, pad_end] specifies the padding for input dimension i + 1, which corresponds to spatial dimension i. It is required that block_shape[i] divides input_shape[i + 1] + pad_start + pad_end.
Returns:
A variable. Has the same type as input.
*/
VARP _SpaceToBatchND(VARP input, VARP block_shape, VARP paddings) {
std::unique_ptr<OpT> op(new OpT);
std::unique_ptr<BlobT> blob_blockShape(new BlobT);
std::unique_ptr<BlobT> blob_paddings(new BlobT);
op->type = OpType_SpaceToBatchND;
auto param = new SpaceBatchT;
auto info_block_shape = block_shape->getInfo();
auto info_paddings = paddings->getInfo();
MNN_ASSERT(info_block_shape != nullptr);
MNN_ASSERT(info_paddings != nullptr);
MNN_ASSERT(halide_type_int == info_block_shape->type.code);
MNN_ASSERT(halide_type_int == info_paddings->type.code);
blob_blockShape->dims = info_block_shape->dim;
blob_blockShape->dataFormat = (MNN::MNN_DATA_FORMAT)Utils::convertFormat(info_block_shape->order);
blob_blockShape->dataType = (MNN::DataType)Utils::convertDataType(info_block_shape->type);
auto data_block_shape = block_shape->readMap<int>();
for (int i=0; i<info_block_shape->size; i++)
{
blob_blockShape->int32s.emplace_back(data_block_shape[i]);
}
blob_paddings->dims = info_paddings->dim;
blob_paddings->dataFormat = (MNN::MNN_DATA_FORMAT)Utils::convertFormat(info_paddings->order);
blob_paddings->dataType = (MNN::DataType)Utils::convertDataType(info_paddings->type);
auto data_paddings = paddings->readMap<int>();
for (int i=0; i<info_paddings->size; i++)
{
blob_paddings->int32s.emplace_back(data_paddings[i]);
}
param->blockShape = std::move(blob_blockShape);
param->padding = std::move(blob_paddings);
op->main.type = OpParameter_SpaceBatch;
op->main.value = param;
return Variable::create(Expr::create(std::move(op), {input}));
}
/*Creates a variable with all elements set to zero.
Args:
input: A variable.
Returns:
A variable with all elements set to zero.
*/
VARP _ZerosLike(VARP input) {
std::unique_ptr<OpT> op(new OpT);
op->type = OpType_ZerosLike;
op->main.type = OpParameter_NONE;
op->main.value = nullptr;
return Variable::create(Expr::create(std::move(op), {input}));
}
/*Unpacks the given dimension of a rank-R tensor into rank-(R-1) variable.
For example, given a variable of shape (A, B, C, D);
If axis == 0 then the i'th variable in output is the slice value[i, :, :, :] and each variable in output will have shape (B, C, D).
(Note that the dimension unpacked along is gone, unlike split).
If axis == 1 then the i'th variable in output is the slice value[:, i, :, :] and each variable in output will have shape (A, C, D).
Args:
value: A rank R > 0 variable to be unstacked.
num: An int. The length of the dimension axis. Automatically inferred if None (the default).
axis: An int. The axis to unstack along. Defaults to the first dimension. Negative values wrap around, so the valid range is [-R, R).
Returns:
The list of variable objects unstacked from value.
*/
std::vector <VARP> _Unstack(VARP value, int axis) {
std::unique_ptr<OpT> op(new OpT);
op->type = OpType_Unpack;
auto info_value = value->getInfo();
MNN_ASSERT(info_value != nullptr);
auto dims = info_value->dim;
auto dimsize = dims.size();
MNN_ASSERT(dimsize > 1);
axis = axis % dimsize;
if(axis < 0) {
axis += dimsize;
}
auto size = dims[axis];
MNN_ASSERT(size > 0);
auto axisParam = new AxisT;
axisParam->axis = axis;
op->main.type = OpParameter_Axis;
op->main.value = axisParam;
EXPRP expr = Expr::create(std::move(op), {value}, size);
std::vector<VARP> res;
for (int i = 0; i < size; ++i) {
res.emplace_back(Variable::create(expr, i));
}
return res;
}
/*Returns the rank of a variable.
Returns a 0-D int32 variable representing the rank of input.
Note: The rank of a variable is not the same as the rank of a matrix.
It's the number of indices required to uniquely select each element of the variable.
It's also known as "order", "degree", or "ndims."
Args:
input: A variable.
Returns:
A 0-D variable of type Halide_Type_Int
*/
VARP _Rank(VARP input) {
std::unique_ptr<OpT> op(new OpT);
op->type = OpType_Rank;
op->main.type = OpParameter_NONE;
op->main.value = nullptr;
return Variable::create(Expr::create(std::move(op), {input}));
}
/*Creates a sequence of numbers.
Args:
start: A 0-D variable (scalar).
limit: A 0-D variable (scalar).
delta: A 0-D variable (scalar).
*/
VARP _Range(VARP start, VARP limit, VARP delta) {
std::unique_ptr<OpT> op(new OpT);
op->type = OpType_Range;
auto rangeParam = new RangeT;
rangeParam->Tidx = (MNN::DataType)Utils::convertDataType(start->getInfo()->type);
op->main.type = OpParameter_Range;
op->main.value = rangeParam;
return Variable::create(Expr::create(std::move(op), {start, limit, delta}));
}
/*Rearranges data from depth into blocks of spatial data.
It is the reverse transformation of SpaceToDepth. More specifically,
it outputs a copy of the input variable where values from the depth dimension are moved in spatial blocks to the height and width dimensions.
Args:
input: A variable.
block_size: An int that is >= 2. The size of the spatial block, same as in Space2Depth.
Returns:
A variable. Has the same type as input.
*/
VARP _DepthToSpace(VARP input, int block_size) {
std::unique_ptr<OpT> op(new OpT);
op->type = OpType_DepthToSpace;
auto depthtospaceParam = new DepthSpaceParamT;
depthtospaceParam->blockSize = block_size;
op->main.type = OpParameter_DepthSpaceParam;
op->main.value = depthtospaceParam;
return Variable::create(Expr::create(std::move(op), {input}));
}
/*SSD network's priorbox layer.
Args:
feature: A variable. Contains the feature map. Namely bottom[0] in caffe.
image: A variable. Contains the image. Namely bottom[1] in caffe.
min_size: Minimum box size (in pixels).
max_size: Maximum box size (in pixels).
aspect_ratio: Various of aspect ratios. Duplicate ratios are ignored. If none is provided, use default 1.0.
flip: If true, flips each aspect ratio. For example, if there is aspect ratio "r", generates aspect ratio "1.0/r" as well. Default true.
clip: If true, clips the prior so that it is within [0, 1]. Default false.
variance: Variance for adjusting the prior bboxes.
img_h: image height. If 0, uses information in image.
img_w: image width. If 0, uses information in image.
step_h: step in height.
step_w: step in width.
offset: Offset to the top left corner of each cell.
Returns:
A variable.
*/
VARP _PriorBox(VARP feature, VARP image, std::vector<float> min_size, std::vector<float> max_size, std::vector<float>aspect_ratio,
bool flip, bool clip, std::vector<float>variance,
unsigned int img_h, unsigned int img_w, float step_h, float step_w, float offset) {
std::unique_ptr<OpT> op(new OpT);
op->type = OpType_PriorBox;
auto param = new PriorBoxT;
param->minSizes = min_size;
param->maxSizes = max_size;
param->aspectRatios = aspect_ratio;
param->flip = flip;
param->clip = clip;
param->variances = variance;
param->imageHeight = img_h;
param->imageWidth = img_w;
param->stepHeight = step_h;
param->stepWidth = step_w;
param->offset = offset;
op->main.type = OpParameter_PriorBox;
op->main.value = param;
return Variable::create(Expr::create(std::move(op), {feature, image}));
}
/*SSD network's permute layer.
Args:
input: A variable. Contains the feature map. Namely bottom[0] in caffe.
dims: A vector. Contains the order.
Returns:
A variable.
*/
VARP _Permute(VARP input, INTS dims) {
std::unique_ptr<OpT> op(new OpT);
op->type = OpType_Permute;
auto param = new PermuteT;
param->dims = dims;
op->main.type = OpParameter_Permute;
op->main.value = param;
return Variable::create(Expr::create(std::move(op), {input}));
}
/*SSD network's detectionoutput layer.
Args:
location: A variable.
confidence: A variable.
priorbox: A variable.
num_classes: number of classes.
share_location: indicates wheter share location between different classes, default true.
background_label_id: default = 0.
nms_threshhold: nonmaximumsupression threshhold.
mns_topk: nonmaximumsupression topk.
code_type: indicates the mode to encode bbox, default = CORNER.
variance_encoded_in_target: indicates whether encode variance in target, default false.
keep_top_k: indicates the number of boxes kept, default -1(all boxes are kept).
confidence_threshold: the threshhold for confidence.
visualize_threshold: The threshold used to visualize the detection results.
Returns:
A variable.
*/
VARP _DetectionOutput(VARP location, VARP confidence, VARP priorbox,
unsigned int num_classes, bool share_location, int background_label_id,
float nms_threshhold, int nms_topk, int code_type,
bool variance_encoded_in_target,
int keep_top_k, float confidence_threshold, float visualize_threshold){
std::unique_ptr<OpT> op(new OpT);
op->type = OpType_DetectionOutput;
auto param = new DetectionOutputT;
param->classCount = num_classes;
param->shareLocation = share_location;
param->backgroundLable = background_label_id;
param->nmsThresholdold = nms_threshhold;
param->nmsTopK = nms_topk;
param->codeType = code_type;
param->varianceEncodedTarget = variance_encoded_in_target;
param->keepTopK = keep_top_k;
param->confidenceThreshold = confidence_threshold;
param->objectnessScore = visualize_threshold;
op->main.type = OpParameter_DetectionOutput;
op->main.value = param;
return Variable::create(Expr::create(std::move(op), {location, confidence, priorbox}));
}
VARP _Interp(VARPS xs, float widthScale, float heightScale, int outputWidth, int outputHeight, int resizeType, bool alignCorners) {
std::unique_ptr<OpT> interp(new OpT);
interp->type = OpType_Interp;
auto param = new InterpT;
param->widthScale = widthScale;
param->heightScale = heightScale;
param->outputWidth = outputWidth;
param->outputHeight = outputHeight;
param->resizeType = resizeType;
param->alignCorners = alignCorners;
interp->main.value = param;
interp->main.type = OpParameter_Interp;
return Variable::create(Expr::create(std::move(interp), xs));
}
VARP _ZeroGrad(VARP x) {
std::unique_ptr<OpT> op(new OpT);
op->type = OpType_ZeroGrad;
return Variable::create(Expr::create(std::move(op), {x}));
}
VARP _Conv(std::vector<int8_t>&& weight, std::vector<int>&& bias, std::vector<float>&& scale, VARP x, INTS channel, INTS kernelSize,
PaddingMode pad, INTS stride, INTS dilate, int group, INTS pads, bool relu) {
std::unique_ptr<OpT> convOp(new OpT);
convOp->type = OpType_ConvInt8;
if (channel[0] == channel[1] && channel[0] == group) {
convOp->type = OpType_DepthwiseConvInt8;
}
convOp->main.type = OpParameter_Convolution2D;
convOp->main.value = new Convolution2DT;
auto conv2D = convOp->main.AsConvolution2D();
conv2D->common.reset(new Convolution2DCommonT);
conv2D->common->padMode = _convertPadMode(pad);
conv2D->common->padX = pads[0];
conv2D->common->padY = pads[1];
conv2D->common->strideX = stride[0];
conv2D->common->strideY = stride[1];
conv2D->common->group = group;
conv2D->common->outputCount = channel[1];
conv2D->common->inputCount = channel[0];
conv2D->common->dilateX = dilate[0];
conv2D->common->dilateY = dilate[1];
conv2D->common->kernelX = kernelSize[0];
conv2D->common->kernelY = kernelSize[1];
conv2D->common->relu = relu;
MNN_ASSERT(weight.size() == channel[1] * (channel[0] / group) * kernelSize[0] * kernelSize[1]);
conv2D->symmetricQuan.reset(new QuantizedFloatParamT);
conv2D->symmetricQuan->bias = std::move(bias);
conv2D->symmetricQuan->scale = std::move(scale);
conv2D->symmetricQuan->weight = std::move(weight);
return (Variable::create(Expr::create(convOp.get(), {x})));
}
VARP _FloatToInt8(VARP x, VARP scale, char minValue/*For future*/, char maxValue/*For future*/) {
auto xInfo = x->getInfo();
auto scaleInfo = scale->getInfo();
auto scalePtr = scale->readMap<float>();
if (nullptr == scalePtr || nullptr == xInfo || nullptr == scaleInfo) {
MNN_ERROR("Error for FloatToInt8 because var not ready\n");
return nullptr;
}
if (xInfo->order != NC4HW4 || xInfo->type.code != halide_type_float || xInfo->dim.size() < 4) {
MNN_ERROR("Not Support Input for FloatToInt8 because var not NC4HW4 or not float\n");
return nullptr;
}
if (scaleInfo->size != xInfo->dim[1]) {
MNN_ERROR("Scale's size not match input's channel: %d - %d\n", scaleInfo->size, xInfo->dim[1]);
return nullptr;
}
std::unique_ptr<OpT> op(new OpT);
op->type = OpType_FloatToInt8;
op->main.type = OpParameter_QuantizedFloatParam;
op->main.value = new QuantizedFloatParamT;
op->main.AsQuantizedFloatParam()->tensorScale.resize(scaleInfo->size);
::memcpy(op->main.AsQuantizedFloatParam()->tensorScale.data(), scalePtr, scaleInfo->size * sizeof(float));
return Variable::create(Expr::create(op.get(), {x}));
}
VARP _Int8ToFloat(VARP x, VARP scale) {
auto xInfo = x->getInfo();
auto scaleInfo = scale->getInfo();
auto scalePtr = scale->readMap<float>();
if (nullptr == scalePtr || nullptr == xInfo || nullptr == scaleInfo) {
MNN_ERROR("Error for _Int8ToFloat because var not ready\n");
return nullptr;
}
if (xInfo->order != NC4HW4 || xInfo->type.code != halide_type_int) {
MNN_ERROR("Not Support Input for _Int8ToFloat because var not NC4HW4 or not int8\n");
return nullptr;
}
if (scaleInfo->size != xInfo->dim[1]) {
MNN_ERROR("_Int8ToFloat Scale's size not match input's channel\n");
return nullptr;
}
std::unique_ptr<OpT> op(new OpT);
op->type = OpType_Int8ToFloat;
op->main.type = OpParameter_QuantizedFloatParam;
op->main.value = new QuantizedFloatParamT;
op->main.AsQuantizedFloatParam()->tensorScale.resize(scaleInfo->size);
::memcpy(op->main.AsQuantizedFloatParam()->tensorScale.data(), scalePtr, scaleInfo->size * sizeof(float));
return Variable::create(Expr::create(op.get(), {x}));
}
} // namespace Express
} // namespace MNN
Reference in New Issue View Git Blame Copy Permalink