You can not select more than 25 topics Topics must start with a chinese character,a letter or number, can include dashes ('-') and can be up to 35 characters long.

test_deconvolutiondepthwise1d.cpp 4.2 kB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115
  1. // Tencent is pleased to support the open source community by making ncnn available.
  2. //
  3. // Copyright (C) 2022 THL A29 Limited, a Tencent company. All rights reserved.
  4. //
  5. // Licensed under the BSD 3-Clause License (the "License"); you may not use this file except
  6. // in compliance with the License. You may obtain a copy of the License at
  7. //
  8. // https://opensource.org/licenses/BSD-3-Clause
  9. //
  10. // Unless required by applicable law or agreed to in writing, software distributed
  11. // under the License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR
  12. // CONDITIONS OF ANY KIND, either express or implied. See the License for the
  13. // specific language governing permissions and limitations under the License.
  14. #include "layer/deconvolutiondepthwise1d.h"
  15. #include "testutil.h"
  16. static int test_deconvolutiondepthwise1d(int w, int h, int outh, int kernel, int dilation, int stride, int pad, int bias, int group, int output_pad_right, int output_w)
  17. {
  18. ncnn::Mat a = RandomMat(w, h);
  19. if (output_w > 0 && pad != -233 && pad != -234)
  20. {
  21. pad = -233;
  22. }
  23. ncnn::ParamDict pd;
  24. pd.set(0, outh);
  25. pd.set(1, kernel);
  26. pd.set(2, dilation);
  27. pd.set(3, stride);
  28. pd.set(4, pad);
  29. pd.set(5, bias);
  30. pd.set(6, outh / group * h / group * kernel * group);
  31. pd.set(7, group);
  32. int activation_type = RAND() % 5; // 0 1 2 3 4
  33. ncnn::Mat activation_params(2);
  34. activation_params[0] = RandomFloat(-1, 0); // alpha
  35. activation_params[1] = RandomFloat(0, 1); // beta
  36. pd.set(9, activation_type);
  37. pd.set(10, activation_params);
  38. pd.set(18, output_pad_right);
  39. pd.set(20, output_w);
  40. std::vector<ncnn::Mat> weights(2);
  41. weights[0] = RandomMat(outh / group * h / group * kernel * group);
  42. weights[1] = RandomMat(outh);
  43. int ret = test_layer<ncnn::DeconvolutionDepthWise1D>("DeconvolutionDepthWise1D", pd, weights, a);
  44. if (ret != 0)
  45. {
  46. fprintf(stderr, "test_deconvolutiondepthwise1d failed w=%d h=%d outh=%d kernel=%d dilation=%d stride=%d pad=%d bias=%d group=%d act=%d actparams=[%f,%f] output_pad_right=%d output_w=%d\n", w, h, outh, kernel, dilation, stride, pad, bias, group, activation_type, activation_params[0], activation_params[1], output_pad_right, output_w);
  47. }
  48. return ret;
  49. }
  50. static int test_deconvolutiondepthwise1d_0()
  51. {
  52. static const int kdsp[16][4] = {
  53. {1, 1, 1, 0},
  54. {1, 1, 2, 0},
  55. {2, 1, 1, 1},
  56. {2, 1, 2, -233},
  57. {3, 1, 1, 1},
  58. {3, 1, 2, 1},
  59. {3, 2, 1, 1},
  60. {4, 1, 1, -233},
  61. {4, 1, 2, -234},
  62. {4, 2, 1, -234},
  63. {5, 1, 1, 2},
  64. {5, 1, 2, 2},
  65. {5, 2, 2, 2},
  66. {7, 1, 1, 3},
  67. {7, 1, 2, 3},
  68. {7, 2, 1, -233},
  69. };
  70. for (int i = 0; i < 16; i++)
  71. {
  72. const int k = kdsp[i][0];
  73. const int d = kdsp[i][1];
  74. const int s = kdsp[i][2];
  75. const int p = kdsp[i][3];
  76. int ret = 0
  77. || test_deconvolutiondepthwise1d(15, 1, 1, k, d, s, p, 1, 1, 0, 0)
  78. || test_deconvolutiondepthwise1d(15, 2, 2, k, d, s, p, 0, 1, 1, 7)
  79. || test_deconvolutiondepthwise1d(15, 2, 2, k, d, s, p, 1, 2, 1, 0)
  80. || test_deconvolutiondepthwise1d(15, 3, 3, k, d, s, p, 0, 3, 0, 0)
  81. || test_deconvolutiondepthwise1d(15, 4, 2, k, d, s, p, 1, 2, 0, 7)
  82. || test_deconvolutiondepthwise1d(15, 4, 4, k, d, s, p, 0, 4, 2, 0)
  83. || test_deconvolutiondepthwise1d(15, 7, 7, k, d, s, p, 1, 7, 2, 0)
  84. || test_deconvolutiondepthwise1d(15, 8, 8, k, d, s, p, 0, 2, 0, 7)
  85. || test_deconvolutiondepthwise1d(15, 8, 8, k, d, s, p, 1, 8, 0, 0)
  86. || test_deconvolutiondepthwise1d(15, 12, 12, k, d, s, p, 0, 4, 3, 0)
  87. || test_deconvolutiondepthwise1d(15, 15, 15, k, d, s, p, 1, 15, 3, 7)
  88. || test_deconvolutiondepthwise1d(15, 16, 8, k, d, s, p, 0, 2, 0, 0)
  89. || test_deconvolutiondepthwise1d(15, 16, 16, k, d, s, p, 1, 16, 0, 0);
  90. if (ret != 0)
  91. return -1;
  92. }
  93. return 0;
  94. }
  95. int main()
  96. {
  97. SRAND(7767517);
  98. return test_deconvolutiondepthwise1d_0();
  99. }