mirror of https://github.com/alibaba/MNN.git
44 lines
1.4 KiB
C++
44 lines
1.4 KiB
C++
//
|
|
// ShapeCrop.cpp
|
|
// MNN
|
|
//
|
|
// Created by MNN on 2019/01/10.
|
|
// Copyright © 2018, Alibaba Group Holding Limited
|
|
//
|
|
|
|
#include "shape/SizeComputer.hpp"
|
|
#include "core/Macro.h"
|
|
|
|
namespace MNN {
|
|
class CropSizeComputer : public SizeComputer {
|
|
virtual bool onComputeSize(const MNN::Op* op, const std::vector<Tensor*>& inputs,
|
|
const std::vector<Tensor*>& outputs) const override {
|
|
MNN_ASSERT(2 == inputs.size());
|
|
MNN_ASSERT(1 == outputs.size());
|
|
MNN_ASSERT(4 == inputs[0]->buffer().dimensions && 4 == inputs[1]->buffer().dimensions);
|
|
MNN_ASSERT(inputs[0]->buffer().dimensions == inputs[1]->buffer().dimensions);
|
|
|
|
auto& ibInput0 = inputs[0]->buffer();
|
|
auto& ibInput1 = inputs[1]->buffer();
|
|
auto& ob = outputs[0]->buffer();
|
|
|
|
ob.dimensions = ibInput1.dimensions;
|
|
::memcpy(ob.dim, ibInput1.dim, ibInput1.dimensions * sizeof(halide_dimension_t));
|
|
|
|
auto cropParam = op->main_as_Crop();
|
|
for (int i = 0; i < ibInput1.dimensions; ++i) {
|
|
if (i < cropParam->axis()) {
|
|
ob.dim[i].extent = ibInput0.dim[i].extent;
|
|
}
|
|
}
|
|
ob.type = ibInput0.type;
|
|
TensorUtils::getDescribe(outputs[0])->dimensionFormat = TensorUtils::getDescribe(inputs[0])->dimensionFormat;
|
|
|
|
return true;
|
|
}
|
|
};
|
|
|
|
REGISTER_SHAPE(CropSizeComputer, OpType_Crop);
|
|
|
|
} // namespace MNN
|