mirror of https://github.com/alibaba/MNN.git
45 lines
1.6 KiB
C++
45 lines
1.6 KiB
C++
//
|
|
// ShapeBatchToSpaceND.cpp
|
|
// MNN
|
|
//
|
|
// Created by MNN on 2019/01/10.
|
|
// Copyright © 2018, Alibaba Group Holding Limited
|
|
//
|
|
|
|
#include "core/SizeComputer.hpp"
|
|
namespace MNN {
|
|
class BatchToSpaceNDSizeComputer : public SizeComputer {
|
|
public:
|
|
virtual bool onComputeSize(const MNN::Op* op, const std::vector<Tensor*>& inputs,
|
|
const std::vector<Tensor*>& outputs) const override {
|
|
auto input = inputs[0];
|
|
auto output = outputs[0];
|
|
|
|
auto paramter = op->main_as_SpaceBatch();
|
|
const auto blockShape = paramter->blockShape();
|
|
const auto blockShapeData = blockShape->int32s()->data();
|
|
int batch = input->batch();
|
|
for (int i = 0; i < blockShape->dims()->data()[0]; ++i) {
|
|
batch /= blockShapeData[i];
|
|
}
|
|
|
|
const auto crop = paramter->padding();
|
|
const auto cropData = crop->int32s()->data();
|
|
int outputHeight = input->height() * blockShapeData[0] - cropData[0] - cropData[1];
|
|
int outputWidth = input->width() * blockShapeData[1] - cropData[2] - cropData[3];
|
|
|
|
output->buffer().dimensions = input->buffer().dimensions;
|
|
output->setLength(0, batch);
|
|
output->setLength(1, input->channel());
|
|
output->setLength(2, outputHeight);
|
|
output->setLength(3, outputWidth);
|
|
output->buffer().type = input->getType();
|
|
TensorUtils::getDescribe(output)->dimensionFormat = MNN_DATA_FORMAT_NC4HW4;
|
|
return true;
|
|
}
|
|
};
|
|
|
|
REGISTER_SHAPE(BatchToSpaceNDSizeComputer, OpType_BatchToSpaceND);
|
|
|
|
} // namespace MNN
|