mirror of https://github.com/alibaba/MNN.git
38 lines
1.4 KiB
C++
38 lines
1.4 KiB
C++
|
//
|
|||
|
|
// ShapeDetectionOutput.cpp
|
|||
|
|
// MNN
|
|||
|
|
//
|
|||
|
|
// Created by MNN on 2019/01/10.
|
|||
|
|
// Copyright © 2018, Alibaba Group Holding Limited
|
|||
|
|
//
|
|||
|
|
|
|||
|
|
#include "Macro.h"
|
|||
|
|
#include "SizeComputer.hpp"
|
|||
|
|
namespace MNN {
|
|||
|
|
|
|||
|
|
// –––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
|
|||
|
|
// Size Computer
|
|||
|
|
// –––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
|
|||
|
|
class DetectionOutputComputer : public SizeComputer {
|
|||
|
|
virtual bool onComputeSize(const MNN::Op *op, const std::vector<Tensor *> &inputs,
|
|||
|
|
const std::vector<Tensor *> &outputs) const override {
|
|||
|
|
MNN_ASSERT(3 <= inputs.size());
|
|||
|
|
MNN_ASSERT(1 == outputs.size());
|
|||
|
|
|
|||
|
|
// set dims
|
|||
|
|
auto &priorbox = inputs[2]->buffer();
|
|||
|
|
auto &output = outputs[0]->buffer();
|
|||
|
|
auto priorCount = priorbox.dim[2].extent / 4;
|
|||
|
|
|
|||
|
|
output.dim[0].extent = 1;
|
|||
|
|
output.dim[1].extent = 1;
|
|||
|
|
output.dim[2].extent = priorCount;
|
|||
|
|
output.dim[3].extent = 6; // maximum width
|
|||
|
|
|
|||
|
|
return true;
|
|||
|
|
}
|
|||
|
|
};
|
|||
|
|
|
|||
|
|
REGISTER_SHAPE(DetectionOutputComputer, OpType_DetectionOutput);
|
|||
|
|
} // namespace MNN
|