-
Notifications
You must be signed in to change notification settings - Fork 212
/
Copy pathyolov8_pose.cpp
214 lines (198 loc) · 11.4 KB
/
yolov8_pose.cpp
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
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
#include"yolov8_pose.h"
#include"decode_yolov8_pose.h"
YOLOv8Pose::YOLOv8Pose(const utils::InitParameter& param) :yolo::YOLO(param),
m_nkpts(17)
{
m_skeleton = { cv::Point2i(16, 14), cv::Point2i(14, 12), cv::Point2i(17, 15),
cv::Point2i(15, 13), cv::Point2i(12, 13), cv::Point2i(6, 12),
cv::Point2i(7, 13), cv::Point2i(6, 7), cv::Point2i(6, 8),
cv::Point2i(7, 9), cv::Point2i(8, 10), cv::Point2i(9, 11),
cv::Point2i(2, 3), cv::Point2i(1, 2), cv::Point2i(1, 3),
cv::Point2i(2, 4), cv::Point2i(3, 5), cv::Point2i(4, 6),
cv::Point2i(5, 7) };
m_kpt_color = { cv::Scalar(0, 255, 0), cv::Scalar(0, 255, 0), cv::Scalar(0, 255, 0),
cv::Scalar(0, 255, 0), cv::Scalar(0, 255, 0), cv::Scalar(255, 128, 0),
cv::Scalar(255, 128, 0), cv::Scalar(255, 128, 0), cv::Scalar(255, 128, 0),
cv::Scalar(255, 128, 0), cv::Scalar(255, 128, 0), cv::Scalar(51, 153, 255),
cv::Scalar(51, 153, 255), cv::Scalar(51, 153, 255), cv::Scalar(51, 153, 255),
cv::Scalar(51, 153, 255), cv::Scalar(51, 153, 255) };
m_limb_color = { cv::Scalar(51, 153, 255), cv::Scalar(51, 153, 255), cv::Scalar(51, 153, 255),
cv::Scalar(51, 153, 255), cv::Scalar(255, 51, 255),cv::Scalar(255, 51, 255),
cv::Scalar(255, 51, 255),cv::Scalar(255, 128, 0), cv::Scalar(255, 128, 0),
cv::Scalar(255, 128, 0), cv::Scalar(255, 128, 0), cv::Scalar(255, 128, 0),
cv::Scalar(0, 255, 0), cv::Scalar(0, 255, 0), cv::Scalar(0, 255, 0),
cv::Scalar(0, 255, 0), cv::Scalar(0, 255, 0), cv::Scalar(0, 255, 0),
cv::Scalar(0, 255, 0) };
m_output_objects_device = nullptr;
m_output_objects_width = 58; // xywhc + points * 17 = 56 -> left, top, right, bottom, confidence, class, keepflag + points * 17 = 58
int output_objects_size = param.batch_size * (1 + param.topK * m_output_objects_width); // 1: count
CHECK(cudaMalloc(&m_output_objects_device, output_objects_size * sizeof(float)));
m_output_objects_host = new float[output_objects_size];
m_objectss.resize(param.batch_size);
}
YOLOv8Pose::~YOLOv8Pose()
{
CHECK(cudaFree(m_output_objects_device));
CHECK(cudaFree(m_output_src_device));
CHECK(cudaFree(m_output_src_transpose_device));
delete[] m_output_objects_host;
m_output_src_device = nullptr;
}
bool YOLOv8Pose::init(const std::vector<unsigned char>& trtFile)
{
if (trtFile.empty())
{
return false;
}
std::unique_ptr<nvinfer1::IRuntime> runtime =
std::unique_ptr<nvinfer1::IRuntime>(nvinfer1::createInferRuntime(sample::gLogger.getTRTLogger()));
if (runtime == nullptr)
{
return false;
}
this->m_engine = std::unique_ptr<nvinfer1::ICudaEngine>(runtime->deserializeCudaEngine(trtFile.data(), trtFile.size()));
if (this->m_engine == nullptr)
{
return false;
}
this->m_context = std::unique_ptr<nvinfer1::IExecutionContext>(this->m_engine->createExecutionContext());
if (this->m_context == nullptr)
{
return false;
}
if (m_param.dynamic_batch)
{
this->m_context->setBindingDimensions(0, nvinfer1::Dims4(m_param.batch_size, 3, m_param.dst_h, m_param.dst_w));
}
m_output_dims = this->m_context->getBindingDimensions(1);
m_total_objects = m_output_dims.d[2];
assert(m_param.batch_size <= m_output_dims.d[0]);
m_output_area = 1;
for (int i = 1; i < m_output_dims.nbDims; i++)
{
if (m_output_dims.d[i] != 0)
{
m_output_area *= m_output_dims.d[i];
}
}
CHECK(cudaMalloc(&m_output_src_device, m_param.batch_size * m_output_area * sizeof(float)));
CHECK(cudaMalloc(&m_output_src_transpose_device, m_param.batch_size * m_output_area * sizeof(float)));
float a = float(m_param.dst_h) / m_param.src_h;
float b = float(m_param.dst_w) / m_param.src_w;
float scale = a < b ? a : b;
cv::Mat src2dst = (cv::Mat_<float>(2, 3) << scale, 0.f, (-scale * m_param.src_w + m_param.dst_w + scale - 1) * 0.5,
0.f, scale, (-scale * m_param.src_h + m_param.dst_h + scale - 1) * 0.5);
cv::Mat dst2src = cv::Mat::zeros(2, 3, CV_32FC1);
cv::invertAffineTransform(src2dst, dst2src);
m_dst2src.v0 = dst2src.ptr<float>(0)[0];
m_dst2src.v1 = dst2src.ptr<float>(0)[1];
m_dst2src.v2 = dst2src.ptr<float>(0)[2];
m_dst2src.v3 = dst2src.ptr<float>(1)[0];
m_dst2src.v4 = dst2src.ptr<float>(1)[1];
m_dst2src.v5 = dst2src.ptr<float>(1)[2];
return true;
}
void YOLOv8Pose::preprocess(const std::vector<cv::Mat>& imgsBatch)
{
resizeDevice(m_param.batch_size, m_input_src_device, m_param.src_w, m_param.src_h,
m_input_resize_device, m_param.dst_w, m_param.dst_h, 114, m_dst2src);
bgr2rgbDevice(m_param.batch_size, m_input_resize_device, m_param.dst_w, m_param.dst_h,
m_input_rgb_device, m_param.dst_w, m_param.dst_h);
normDevice(m_param.batch_size, m_input_rgb_device, m_param.dst_w, m_param.dst_h,
m_input_norm_device, m_param.dst_w, m_param.dst_h, m_param);
hwc2chwDevice(m_param.batch_size, m_input_norm_device, m_param.dst_w, m_param.dst_h,
m_input_hwc_device, m_param.dst_w, m_param.dst_h);
}
void YOLOv8Pose::postprocess(const std::vector<cv::Mat>& imgsBatch)
{
yolov8pose::transposeDevice(m_param, m_output_src_device, m_total_objects, 56, m_total_objects * 56,
m_output_src_transpose_device, 56, m_total_objects);
yolov8pose::decodeDevice(m_param, m_output_src_transpose_device, 56, m_total_objects, m_output_area,
m_output_objects_device, m_output_objects_width, m_param.topK);
nmsDeviceV1(m_param, m_output_objects_device, m_output_objects_width, m_param.topK, m_param.topK * m_output_objects_width + 1);
//nmsDeviceV2(m_param, m_output_objects_device, m_output_objects_width, m_param.topK, m_param.topK * m_output_objects_width + 1, m_output_idx_device, m_output_conf_device);
CHECK(cudaMemcpy(m_output_objects_host, m_output_objects_device, m_param.batch_size * sizeof(float) * (1 + m_output_objects_width * m_param.topK), cudaMemcpyDeviceToHost));
}
void YOLOv8Pose::reset()
{
CHECK(cudaMemset(m_output_objects_device, 0, sizeof(float) * m_param.batch_size * (1 + m_output_objects_width * m_param.topK)));
}
void YOLOv8Pose::showAndSave(const std::vector<std::string>& classNames, const int& cvDelayTime, std::vector<cv::Mat>& imgsBatch, float* avg_times)
{
if (!m_param.is_show && !m_param.is_save)
return;
cv::Scalar color = cv::Scalar(0, 255, 0);
cv::Point bbox_points[1][4];
const cv::Point* bbox_point0[1] = { bbox_points[0] };
int num_points[] = { 4 };
for (size_t bi = 0; bi < imgsBatch.size(); bi++)
{
int num_boxes = std::min((int)(m_output_objects_host + bi * (m_param.topK * m_output_objects_width + 1))[0], m_param.topK);
for (size_t i = 0; i < num_boxes; i++)
{
float* ptr = m_output_objects_host + bi * (m_param.topK * m_output_objects_width + 1) + m_output_objects_width * i + 1;
int keep_flag = ptr[6];
if (keep_flag)
{
int label = (int)ptr[5];
color = utils::Colors::color80[label];
float x_lt = m_dst2src.v0 * ptr[0] + m_dst2src.v1 * ptr[1] + m_dst2src.v2;
float y_lt = m_dst2src.v3 * ptr[0] + m_dst2src.v4 * ptr[1] + m_dst2src.v5;
float x_rb = m_dst2src.v0 * ptr[2] + m_dst2src.v1 * ptr[3] + m_dst2src.v2;
float y_rb = m_dst2src.v3 * ptr[2] + m_dst2src.v4 * ptr[3] + m_dst2src.v5;
cv::rectangle(imgsBatch[bi], cv::Point(x_lt, y_lt), cv::Point(x_rb, y_rb), color, 2, cv::LINE_AA);
cv::String det_info = m_param.class_names[label] + " " + cv::format("%.4f", ptr[4]);
bbox_points[0][0] = cv::Point(x_lt, y_lt);
bbox_points[0][1] = cv::Point(x_lt + det_info.size() * m_param.char_width, y_lt);
bbox_points[0][2] = cv::Point(x_lt + det_info.size() * m_param.char_width, y_lt - m_param.det_info_render_width);
bbox_points[0][3] = cv::Point(x_lt, y_lt - m_param.det_info_render_width);
cv::fillPoly(imgsBatch[bi], bbox_point0, num_points, 1, color);
cv::putText(imgsBatch[bi], det_info, bbox_points[0][0], cv::FONT_HERSHEY_DUPLEX, m_param.font_scale, cv::Scalar(255, 255, 255), 1, cv::LINE_AA);
cv::rectangle(imgsBatch[bi], cv::Point(x_lt, y_lt), cv::Point(x_rb, y_rb), color, 1, cv::LINE_AA);
float* pkpt = ptr + 7;
for (size_t pi = 0; pi < m_nkpts; pi++)
{
float conf = pkpt[pi * 3 + 2];
if (conf < 0.5f)
continue;
pkpt[pi * 3] = m_dst2src.v0 * pkpt[pi * 3] + m_dst2src.v1 * pkpt[pi * 3 + 1] + m_dst2src.v2;
pkpt[pi * 3 + 1] = m_dst2src.v3 * pkpt[pi * 3] + m_dst2src.v4 * pkpt[pi * 3 + 1] + m_dst2src.v5;
if (pkpt[pi * 3] >= (float)m_param.src_w || pkpt[pi * 3] < 0 ||
pkpt[pi * 3 + 1] >= (float)m_param.src_h || pkpt[pi * 3 + 1] < 0)
continue;
cv::circle(imgsBatch[bi], cv::Size2i(int(pkpt[pi * 3 + 0]), int(pkpt[pi * 3 + 1])), 5, m_kpt_color[pi], -1, cv::LINE_AA);
}
for (size_t ki = 0; ki < m_skeleton.size(); ki++)
{
float conf1 = pkpt[(m_skeleton[ki].x - 1) * 3 + 2];
float conf2 = pkpt[(m_skeleton[ki].y - 1) * 3 + 2];
if (conf1 < 0.5f || conf2 < 0.5f)
continue;
int x0 = int(pkpt[(m_skeleton[ki].x - 1) * 3 + 0]);
int y0 = int(pkpt[(m_skeleton[ki].x - 1) * 3 + 1]);
int x1 = int(pkpt[(m_skeleton[ki].y - 1) * 3 + 0]);
int y1 = int(pkpt[(m_skeleton[ki].y - 1) * 3 + 1]);
cv::Point2i pos1(x0, y0);
cv::Point2i pos2(x1, y1);
if (pos1.x % imgsBatch[bi].cols == 0 || pos1.y % imgsBatch[bi].rows == 0 || pos1.x < 0 or pos1.y < 0)
continue;
if (pos2.x % imgsBatch[bi].cols == 0 || pos2.y % imgsBatch[bi].rows == 0 || pos2.x < 0 or pos2.y < 0)
continue;
cv::line(imgsBatch[bi], pos1, pos2, m_limb_color[ki], 2, cv::LINE_AA);
}
}
}
cv::putText(imgsBatch[bi], "preprocess time =" + cv::format("%.3f", avg_times[0]) + "ms", cv::Point(100, 100), cv::FONT_HERSHEY_DUPLEX, m_param.font_scale + 0.3, cv::Scalar(255, 0, 0), 1, cv::LINE_AA);
cv::putText(imgsBatch[bi], "inference time =" + cv::format("%.3f", avg_times[1]) + "ms", cv::Point(100, 135), cv::FONT_HERSHEY_DUPLEX, m_param.font_scale + 0.3, cv::Scalar(0, 255, 0), 1, cv::LINE_AA);
cv::putText(imgsBatch[bi], "postprocess time =" + cv::format("%.3f", avg_times[2]) + "ms", cv::Point(100, 170), cv::FONT_HERSHEY_DUPLEX, m_param.font_scale + 0.3, cv::Scalar(0, 0, 255), 1, cv::LINE_AA);
if (m_param.is_show)
{
cv::imshow(m_param.winname, imgsBatch[bi]);
cv::waitKey(cvDelayTime);
}
if (m_param.is_save)
{
cv::imwrite(m_param.save_path + utils::getTimeStamp() + ".jpg", imgsBatch[bi]);
}
}
}