mindspore2022/mindspore/ccsrc/minddata/dataset/api/transforms.cc

383 lines
15 KiB
C++

/**
* Copyright 2020 Huawei Technologies Co., Ltd
*
* Licensed under the Apache License, Version 2.0 (the "License");
* you may not use this file except in compliance with the License.
* You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*/
#include "minddata/dataset/include/transforms.h"
// Kernel data headers (in alphabetical order)
#include "minddata/dataset/kernels/data/compose_op.h"
#include "minddata/dataset/kernels/data/duplicate_op.h"
#include "minddata/dataset/kernels/data/one_hot_op.h"
#include "minddata/dataset/kernels/data/random_apply_op.h"
#include "minddata/dataset/kernels/data/random_choice_op.h"
#include "minddata/dataset/kernels/data/type_cast_op.h"
#ifndef ENABLE_ANDROID
#include "minddata/dataset/kernels/data/unique_op.h"
#endif
namespace mindspore {
namespace dataset {
/* ####################################### Validator Functions ############################################ */
Status ValidateProbability(const std::string &op_name, const float probability) {
if (probability < 0.0 || probability > 1.0) {
std::string err_msg = op_name + ": probability must be between 0.0 and 1.0, got: " + std::to_string(probability);
MS_LOG(ERROR) << err_msg;
RETURN_STATUS_SYNTAX_ERROR(err_msg);
}
return Status::OK();
}
Status ValidateIntScalarPositive(const std::string &op_name, const std::string &scalar_name, int32_t scalar) {
RETURN_IF_NOT_OK(ValidateScalar(op_name, scalar_name, scalar, {0}, true));
return Status::OK();
}
Status ValidateFloatScalarPositive(const std::string &op_name, const std::string &scalar_name, float scalar) {
RETURN_IF_NOT_OK(ValidateScalar(op_name, scalar_name, scalar, {0}, true));
return Status::OK();
}
Status ValidateVectorFillvalue(const std::string &op_name, const std::vector<uint8_t> &fill_value) {
if (fill_value.empty() || (fill_value.size() != 1 && fill_value.size() != 3)) {
std::string err_msg =
op_name + ": fill_value expecting size 1 or 3, got fill_value.size(): " + std::to_string(fill_value.size());
MS_LOG(ERROR) << err_msg;
RETURN_STATUS_SYNTAX_ERROR(err_msg);
}
// Note that fill_value need to be in range [0, 255],
// but we omit the check since its type is uint8_t
return Status::OK();
}
Status ValidateVectorColorAttribute(const std::string &op_name, const std::string &attr_name,
const std::vector<float> &attr, const std::vector<float> &range) {
if (attr.empty() || attr.size() > 2) {
std::string err_msg = op_name + ":" + attr_name + " expecting size 1 or 2, but got: " + std::to_string(attr.size());
MS_LOG(ERROR) << err_msg;
RETURN_STATUS_SYNTAX_ERROR(err_msg);
}
for (auto &attr_val : attr) {
RETURN_IF_NOT_OK(ValidateScalar(op_name, attr_name, attr_val, range, false, false));
}
if (attr.size() == 2 && (attr[0] > attr[1])) {
std::string err_msg = op_name + ":" + attr_name +
" lower bound must be less or equal to upper bound, got lb: " + std::to_string(attr[0]) +
", ub: " + std::to_string(attr[1]);
MS_LOG(ERROR) << err_msg;
RETURN_STATUS_SYNTAX_ERROR(err_msg);
}
return Status::OK();
}
Status ValidateVectorMeanStd(const std::string &op_name, const std::vector<float> &mean,
const std::vector<float> &std) {
if (mean.size() != 3) {
std::string err_msg = op_name + ": mean expecting size 3, got size: " + std::to_string(mean.size());
MS_LOG(ERROR) << err_msg;
RETURN_STATUS_SYNTAX_ERROR(err_msg);
}
if (std.size() != 3) {
std::string err_msg = op_name + ": std expecting size 3, got size: " + std::to_string(std.size());
MS_LOG(ERROR) << err_msg;
RETURN_STATUS_SYNTAX_ERROR(err_msg);
}
// check std/mean value
for (int32_t i = 0; i < std.size(); ++i) {
RETURN_IF_NOT_OK(ValidateScalar(op_name, "mean", mean[i], {0.0, 255.0}, false, false));
RETURN_IF_NOT_OK(ValidateScalar(op_name, "std", std[i], {0.0, 255.0}, true, false));
}
return Status::OK();
}
Status ValidateVectorPadding(const std::string &op_name, const std::vector<int32_t> &padding) {
if (padding.empty() || padding.size() == 3 || padding.size() > 4) {
std::string err_msg = op_name + ": padding expecting size 1, 2 or 4, got size: " + std::to_string(padding.size());
MS_LOG(ERROR) << err_msg;
RETURN_STATUS_SYNTAX_ERROR(err_msg);
}
for (const auto &pad_val : padding) {
RETURN_IF_NOT_OK(ValidateScalar(op_name, "padding", pad_val, {0, INT_MAX}, false, false));
}
return Status::OK();
}
Status ValidateVectorPositive(const std::string &op_name, const std::string &vec_name,
const std::vector<int32_t> &vec) {
for (const auto &vec_val : vec) {
RETURN_IF_NOT_OK(ValidateScalar(op_name, vec_name, vec_val, {0}, true));
}
return Status::OK();
}
Status ValidateVectorNonNegative(const std::string &op_name, const std::string &vec_name,
const std::vector<int32_t> &vec) {
for (const auto &vec_val : vec) {
RETURN_IF_NOT_OK(ValidateScalar(op_name, vec_name, vec_val, {0}, false));
}
return Status::OK();
}
Status ValidateVectorSize(const std::string &op_name, const std::vector<int32_t> &size) {
if (size.empty() || size.size() > 2) {
std::string err_msg = op_name + ": size expecting size 2, got size.size(): " + std::to_string(size.size());
MS_LOG(ERROR) << err_msg;
RETURN_STATUS_SYNTAX_ERROR(err_msg);
}
for (const auto &size_val : size) {
RETURN_IF_NOT_OK(ValidateScalar(op_name, "size", size_val, {0, INT_MAX}, true, false));
}
return Status::OK();
}
Status ValidateVectorScale(const std::string &op_name, const std::vector<float> &scale) {
if (scale.size() != 2) {
std::string err_msg = op_name + ": scale expecting size 2, got scale.size(): " + std::to_string(scale.size());
MS_LOG(ERROR) << err_msg;
RETURN_STATUS_SYNTAX_ERROR(err_msg);
}
RETURN_IF_NOT_OK(ValidateScalar(op_name, "scale", scale[0], {0}, false));
RETURN_IF_NOT_OK(ValidateScalar(op_name, "scale", scale[1], {0}, true));
if (scale[1] < scale[0]) {
std::string err_msg = op_name + ": scale must be in the format of (min, max).";
MS_LOG(ERROR) << op_name + ": scale must be in the format of (min, max), but got: " << scale;
RETURN_STATUS_SYNTAX_ERROR(err_msg);
}
return Status::OK();
}
Status ValidateVectorRatio(const std::string &op_name, const std::vector<float> &ratio) {
if (ratio.size() != 2) {
std::string err_msg = op_name + ": ratio expecting size 2, got ratio.size(): " + std::to_string(ratio.size());
MS_LOG(ERROR) << err_msg;
RETURN_STATUS_SYNTAX_ERROR(err_msg);
}
RETURN_IF_NOT_OK(ValidateScalar(op_name, "scale", ratio[0], {0}, true));
RETURN_IF_NOT_OK(ValidateScalar(op_name, "scale", ratio[1], {0}, true));
if (ratio[1] < ratio[0]) {
std::string err_msg = op_name + ": ratio must be in the format of (min, max).";
MS_LOG(ERROR) << op_name + ": ratio must be in the format of (min, max), but got: " << ratio;
RETURN_STATUS_SYNTAX_ERROR(err_msg);
}
return Status::OK();
}
Status ValidateVectorTransforms(const std::string &op_name,
const std::vector<std::shared_ptr<TensorOperation>> &transforms) {
if (transforms.empty()) {
std::string err_msg = op_name + ": transform list must not be empty.";
MS_LOG(ERROR) << err_msg;
RETURN_STATUS_SYNTAX_ERROR(err_msg);
}
for (int32_t i = 0; i < transforms.size(); ++i) {
if (transforms[i] == nullptr) {
std::string err_msg =
op_name + ": transform ops must not be null, got transform[" + std::to_string(i) + "] == nullptr.";
MS_LOG(ERROR) << err_msg;
RETURN_STATUS_SYNTAX_ERROR(err_msg);
}
}
return Status::OK();
}
bool CmpFloat(const float a, const float b, float epsilon) { return (std::fabs(a - b) < epsilon); }
// Transform operations for data.
namespace transforms {
// FUNCTIONS TO CREATE DATA TRANSFORM OPERATIONS
// (In alphabetical order)
// Function to create ComposeOperation.
std::shared_ptr<ComposeOperation> Compose(const std::vector<std::shared_ptr<TensorOperation>> &transforms) {
auto op = std::make_shared<ComposeOperation>(transforms);
// Input validation
return op->ValidateParams() ? op : nullptr;
}
// Function to create DuplicateOperation.
std::shared_ptr<DuplicateOperation> Duplicate() {
auto op = std::make_shared<DuplicateOperation>();
// Input validation
return op->ValidateParams() ? op : nullptr;
}
// Function to create OneHotOperation.
std::shared_ptr<OneHotOperation> OneHot(int32_t num_classes) {
auto op = std::make_shared<OneHotOperation>(num_classes);
// Input validation
return op->ValidateParams() ? op : nullptr;
}
// Function to create RandomApplyOperation.
std::shared_ptr<RandomApplyOperation> RandomApply(const std::vector<std::shared_ptr<TensorOperation>> &transforms,
double prob) {
auto op = std::make_shared<RandomApplyOperation>(transforms, prob);
// Input validation
return op->ValidateParams() ? op : nullptr;
}
// Function to create RandomChoiceOperation.
std::shared_ptr<RandomChoiceOperation> RandomChoice(const std::vector<std::shared_ptr<TensorOperation>> &transforms) {
auto op = std::make_shared<RandomChoiceOperation>(transforms);
// Input validation
return op->ValidateParams() ? op : nullptr;
}
// Function to create TypeCastOperation.
std::shared_ptr<TypeCastOperation> TypeCast(std::string data_type) {
auto op = std::make_shared<TypeCastOperation>(data_type);
// Input validation
return op->ValidateParams() ? op : nullptr;
}
#ifndef ENABLE_ANDROID
// Function to create UniqueOperation.
std::shared_ptr<UniqueOperation> Unique() {
auto op = std::make_shared<UniqueOperation>();
// Input validation
return op->ValidateParams() ? op : nullptr;
}
#endif
/* ####################################### Validator Functions ############################################ */
/* ####################################### Derived TensorOperation classes ################################# */
// (In alphabetical order)
// ComposeOperation
ComposeOperation::ComposeOperation(const std::vector<std::shared_ptr<TensorOperation>> &transforms)
: transforms_(transforms) {}
Status ComposeOperation::ValidateParams() {
RETURN_IF_NOT_OK(ValidateVectorTransforms("Compose", transforms_));
return Status::OK();
}
std::shared_ptr<TensorOp> ComposeOperation::Build() {
std::vector<std::shared_ptr<TensorOp>> tensor_ops;
(void)std::transform(transforms_.begin(), transforms_.end(), std::back_inserter(tensor_ops),
[](std::shared_ptr<TensorOperation> op) -> std::shared_ptr<TensorOp> { return op->Build(); });
return std::make_shared<ComposeOp>(tensor_ops);
}
// DuplicateOperation
Status DuplicateOperation::ValidateParams() { return Status::OK(); }
std::shared_ptr<TensorOp> DuplicateOperation::Build() { return std::make_shared<DuplicateOp>(); }
// OneHotOperation
OneHotOperation::OneHotOperation(int32_t num_classes) : num_classes_(num_classes) {}
Status OneHotOperation::ValidateParams() {
if (num_classes_ <= 0) {
std::string err_msg = "OneHot: Number of classes must be greater than 0, but got: " + std::to_string(num_classes_);
MS_LOG(ERROR) << err_msg;
RETURN_STATUS_SYNTAX_ERROR(err_msg);
}
return Status::OK();
}
std::shared_ptr<TensorOp> OneHotOperation::Build() { return std::make_shared<OneHotOp>(num_classes_); }
// PreBuiltOperation
PreBuiltOperation::PreBuiltOperation(std::shared_ptr<TensorOp> tensor_op) : op_(tensor_op) {}
Status PreBuiltOperation::ValidateParams() { return Status::OK(); }
std::shared_ptr<TensorOp> PreBuiltOperation::Build() { return op_; }
std::string PreBuiltOperation::Name() const { return op_ ? op_->Name() : kPreBuiltOperation; }
Status PreBuiltOperation::to_json(nlohmann::json *out_json) {
RETURN_IF_NOT_OK(op_->to_json(out_json));
return Status::OK();
}
// RandomApplyOperation
RandomApplyOperation::RandomApplyOperation(const std::vector<std::shared_ptr<TensorOperation>> &transforms, double prob)
: TensorOperation(true), transforms_(transforms), prob_(prob) {}
Status RandomApplyOperation::ValidateParams() {
RETURN_IF_NOT_OK(ValidateVectorTransforms("RandomApply", transforms_));
RETURN_IF_NOT_OK(ValidateProbability("RandomApply", prob_));
return Status::OK();
}
std::shared_ptr<TensorOp> RandomApplyOperation::Build() {
std::vector<std::shared_ptr<TensorOp>> tensor_ops;
(void)std::transform(transforms_.begin(), transforms_.end(), std::back_inserter(tensor_ops),
[](std::shared_ptr<TensorOperation> op) -> std::shared_ptr<TensorOp> { return op->Build(); });
return std::make_shared<RandomApplyOp>(prob_, tensor_ops);
}
// RandomChoiceOperation
RandomChoiceOperation::RandomChoiceOperation(const std::vector<std::shared_ptr<TensorOperation>> &transforms)
: TensorOperation(true), transforms_(transforms) {}
Status RandomChoiceOperation::ValidateParams() {
RETURN_IF_NOT_OK(ValidateVectorTransforms("RandomChoice", transforms_));
return Status::OK();
}
std::shared_ptr<TensorOp> RandomChoiceOperation::Build() {
std::vector<std::shared_ptr<TensorOp>> tensor_ops;
(void)std::transform(transforms_.begin(), transforms_.end(), std::back_inserter(tensor_ops),
[](std::shared_ptr<TensorOperation> op) -> std::shared_ptr<TensorOp> { return op->Build(); });
return std::make_shared<RandomChoiceOp>(tensor_ops);
}
// TypeCastOperation
TypeCastOperation::TypeCastOperation(std::string data_type) : data_type_(data_type) {}
Status TypeCastOperation::ValidateParams() {
std::vector<std::string> predefine_type = {"bool", "int8", "uint8", "int16", "uint16", "int32", "uint32",
"int64", "uint64", "float16", "float32", "float64", "string"};
auto itr = std::find(predefine_type.begin(), predefine_type.end(), data_type_);
if (itr == predefine_type.end()) {
std::string err_msg = "TypeCast: Invalid data type: " + data_type_;
MS_LOG(ERROR) << "TypeCast: Only supports data type bool, int8, uint8, int16, uint16, int32, uint32, "
<< "int64, uint64, float16, float32, float64, string, but got: " << data_type_;
RETURN_STATUS_SYNTAX_ERROR(err_msg);
}
return Status::OK();
}
std::shared_ptr<TensorOp> TypeCastOperation::Build() { return std::make_shared<TypeCastOp>(data_type_); }
#ifndef ENABLE_ANDROID
// UniqueOperation
Status UniqueOperation::ValidateParams() { return Status::OK(); }
std::shared_ptr<TensorOp> UniqueOperation::Build() { return std::make_shared<UniqueOp>(); }
#endif
} // namespace transforms
} // namespace dataset
} // namespace mindspore