1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
|
// ----------------------------------------------------------------------------
// - Open3D: www.open3d.org -
// ----------------------------------------------------------------------------
// Copyright (c) 2018-2024 www.open3d.org
// SPDX-License-Identifier: MIT
// ----------------------------------------------------------------------------
#include "open3d/io/PointCloudIO.h"
#include <benchmark/benchmark.h>
#include "open3d/utility/Logging.h"
namespace open3d {
namespace benchmarks {
using open3d::io::ReadPointCloud;
using open3d::io::WritePointCloud;
namespace {
template <class T>
double AverageDistance(const std::vector<T> &a, const std::vector<T> &b) {
if (a.size() != b.size()) {
utility::LogError("vectors different size {} {}", a.size(), b.size());
}
if (a.size() == 0) {
return 0.;
}
double total = 0;
for (size_t i = 0; i < a.size(); ++i) {
total += (a[i] - b[i]).norm();
}
return total / a.size();
}
enum class IsAscii : bool { BINARY = false, ASCII = true };
enum class Compressed : bool { UNCOMPRESSED = false, COMPRESSED = true };
enum class Compare : uint32_t {
// Points are always compared
NONE = 0,
NORMALS = 1 << 0,
COLORS = 1 << 1,
NORMALS_AND_COLORS = NORMALS | COLORS
};
struct ReadWritePCArgs {
std::string filename;
IsAscii write_ascii;
Compressed compressed;
Compare compare;
};
std::vector<ReadWritePCArgs> g_pc_args({
// PCD has ASCII, BINARY, and BINARY_COMPRESSED
{"testau.pcd", IsAscii::ASCII, Compressed::UNCOMPRESSED,
Compare::NORMALS_AND_COLORS}, // 0
{"testbu.pcd", IsAscii::BINARY, Compressed::UNCOMPRESSED,
Compare::NORMALS_AND_COLORS}, // 1
{"testbc.pcd", IsAscii::BINARY, Compressed::COMPRESSED,
Compare::NORMALS_AND_COLORS}, // 2
{"testb.ply", IsAscii::BINARY, Compressed::UNCOMPRESSED,
Compare::NORMALS_AND_COLORS}, // 3
{"testa.ply", IsAscii::ASCII, Compressed::UNCOMPRESSED,
Compare::NORMALS_AND_COLORS}, // 4
{"test.pts", IsAscii::BINARY, Compressed::UNCOMPRESSED,
Compare::COLORS}, // 5
{"test.xyz", IsAscii::BINARY, Compressed::UNCOMPRESSED,
Compare::NONE}, // 6
{"test.xyzn", IsAscii::BINARY, Compressed::UNCOMPRESSED,
Compare::NORMALS}, // 7
{"test.xyzrgb", IsAscii::BINARY, Compressed::UNCOMPRESSED,
Compare::COLORS}, // 8
});
class TestPCGrid0 {
geometry::PointCloud pc_;
int size_ = 0;
const bool print_progress = false;
public:
void Setup(int size) {
if (size_ == size) return;
utility::LogInfo("setup PCGrid size={}", size);
pc_.Clear();
size_ = size;
for (int i = 0; i < size; ++i) {
// provide somewhat random numbers everywhere, so compression
// doesn't get a free pass
pc_.points_.push_back({std::sin(i * .8969920581) * 1000.,
std::sin(i * .3898546778) * 1000.,
std::sin(i * .2509962463) * 1000.});
pc_.normals_.push_back({std::sin(i * .4472367685),
std::sin(i * .9698787116),
std::sin(i * .7072878517)});
// color needs to be [0,1]
pc_.colors_.push_back({std::fmod(i * .4241490710, 1.0),
std::fmod(i * .6468026221, 1.0),
std::fmod(i * .5376722873, 1.0)});
}
}
void WriteRead(int pc_args_id) {
const auto &args = g_pc_args[pc_args_id];
const auto &pc = pc_;
// we loose some precision when saving generated data
if (!WritePointCloud(args.filename, pc,
{bool(args.write_ascii), bool(args.compressed),
print_progress})) {
utility::LogError("Failed to write to {}", args.filename);
}
geometry::PointCloud pc2;
if (!ReadPointCloud(args.filename, pc2,
{"auto", false, false, print_progress})) {
utility::LogError("Failed to read from {}", args.filename);
}
auto CheckLE = [](double a, double b) {
if (a <= b) return;
utility::LogError("Error too high: {} {}", a, b);
};
const double pointsMaxError =
1e-3; //.ply ascii has the highest error, others <1e-4
CheckLE(AverageDistance(pc.points_, pc2.points_), pointsMaxError);
if (int(args.compare) & int(Compare::NORMALS)) {
const double normalsMaxError =
1e-6; //.ply ascii has the highest error, others <1e-7
CheckLE(AverageDistance(pc.normals_, pc2.normals_),
normalsMaxError);
}
if (int(args.compare) & int(Compare::COLORS)) {
const double colorsMaxError =
1e-2; // colors are saved as uint8_t[3] in a lot of formats
CheckLE(AverageDistance(pc.colors_, pc2.colors_), colorsMaxError);
}
}
};
// reuse the same instance so we don't recreate the point cloud every time
TestPCGrid0 test_pc_grid0;
} // namespace
static void BM_TestPCGrid0(::benchmark::State &state) {
// state.range(n) are arguments that are passed to us
int pc_args_id = state.range(0);
int size = state.range(1);
test_pc_grid0.Setup(size);
for (auto _ : state) {
test_pc_grid0.WriteRead(pc_args_id);
}
}
static void BM_TestPCGrid0_Args(benchmark::internal::Benchmark *b) {
for (int j = 4 * 1024; j <= 256 * 1024; j *= 8) {
for (int i = 0; i < int(g_pc_args.size()); ++i) {
b->Args({i, j});
}
}
}
BENCHMARK(BM_TestPCGrid0)->MinTime(0.1)->Apply(BM_TestPCGrid0_Args);
} // namespace benchmarks
} // namespace open3d
|
