wehub-resource-sync
64 lines
3.0 KiB
C++
64 lines
3.0 KiB
C++
//
|
|
// CropAndResizeTest.cpp
|
|
// MNNTests
|
|
//
|
|
// Created by MNN on 2020/08/05.
|
|
// Copyright © 2018, Alibaba Group Holding Limited
|
|
//
|
|
|
|
#include <MNN/expr/Expr.hpp>
|
|
#include <MNN/expr/ExprCreator.hpp>
|
|
#include "MNNTestSuite.h"
|
|
#include "TestUtils.h"
|
|
|
|
using namespace MNN::Express;
|
|
class CropAndResizeTest : public MNNTestCase {
|
|
public:
|
|
virtual ~CropAndResizeTest() = default;
|
|
virtual bool run(int precision) {
|
|
const int batch = 2, inputHeight = 16, inputWidth = 16, depth = 4, boxNum = 2;
|
|
auto img = _Input({batch, inputHeight, inputWidth, depth}, NHWC);
|
|
auto boxes = _Input({boxNum, 4}, NHWC);
|
|
auto box_ind = _Input({boxNum}, NHWC, halide_type_of<uint32_t>());
|
|
auto crop_size = _Input({2}, NHWC, halide_type_of<uint32_t>());
|
|
auto imgPtr = img->writeMap<float>();
|
|
for (int i = 0; i < batch * inputHeight * inputWidth * depth; i++) {
|
|
imgPtr[i] = static_cast<float>((i % 255) / 255.0f);
|
|
}
|
|
const float box_data[] = {
|
|
0.2, 0.3, 0.4, 0.5, 0.3, 0.4, 0.6, 0.7,
|
|
};
|
|
memcpy(boxes->writeMap<float>(), box_data, boxNum * 4 * sizeof(float));
|
|
box_ind->writeMap<uint32_t>()[0] = 1;
|
|
box_ind->writeMap<uint32_t>()[1] = 0;
|
|
crop_size->writeMap<uint32_t>()[0] = 4;
|
|
crop_size->writeMap<uint32_t>()[1] = 4;
|
|
auto output = _CropAndResize(img, boxes, box_ind, crop_size, BILINEAR);
|
|
const std::vector<float> expectedOutput = {
|
|
0.8392, 0.8431, 0.8471, 0.8510, 0.8549, 0.8588, 0.8627, 0.8667, 0.8706, 0.8745, 0.8784, 0.8824, 0.8863,
|
|
0.8902, 0.8941, 0.8980,
|
|
|
|
0.0902, 0.0941, 0.0980, 0.1020, 0.1059, 0.1098, 0.1137, 0.1176, 0.1216, 0.1255, 0.1294, 0.1333, 0.1373,
|
|
0.1412, 0.1451, 0.1490,
|
|
|
|
0.3412, 0.3451, 0.3490, 0.3529, 0.3569, 0.3608, 0.3647, 0.3686, 0.3725, 0.3765, 0.3804, 0.3843, 0.3882,
|
|
0.3922, 0.3961, 0.4000,
|
|
|
|
0.5922, 0.5961, 0.6000, 0.6039, 0.6078, 0.6118, 0.6157, 0.6196, 0.6235, 0.6275, 0.6314, 0.6353, 0.6392,
|
|
0.6431, 0.6471, 0.6510,
|
|
|
|
0.2235, 0.2275, 0.2314, 0.2353, 0.2471, 0.2510, 0.2549, 0.2588, 0.2706, 0.2745, 0.2784, 0.2824, 0.2941,
|
|
0.2980, 0.3020, 0.3059, 0.6000, 0.6039, 0.6078, 0.6118, 0.6235, 0.6275, 0.6314, 0.6353, 0.6471, 0.6510,
|
|
0.6549, 0.6588, 0.6706, 0.6745, 0.6784, 0.6824, 0.4765, 0.4804, 0.4843, 0.4882, 0.5000, 0.5039, 0.5078,
|
|
0.5118, 0.5235, 0.5275, 0.5314, 0.5353, 0.5471, 0.5510, 0.5549, 0.5588, 0.3529, 0.3569, 0.3608, 0.3647,
|
|
0.3765, 0.3804, 0.3843, 0.3882, 0.4000, 0.4039, 0.4078, 0.4118, 0.4235, 0.4275, 0.4314, 0.4353};
|
|
auto gotOutput = output->readMap<float>();
|
|
if (!checkVector<float>(gotOutput, expectedOutput.data(), expectedOutput.size(), 0.01)) {
|
|
MNN_ERROR("CropAndResizeTest test failed!\n");
|
|
return false;
|
|
}
|
|
return true;
|
|
}
|
|
};
|
|
MNNTestSuiteRegister(CropAndResizeTest, "op/CropAndResize");
|