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
|
#include <iostream>
#include <fstream>
#include <cstdlib>
#include <vector>
#include <string>
#include <iterator>
#include <unordered_map>
#include <algorithm>
#include <algorithm>
#include <functional>
#include <queue>
#include <stdlib.h>
#include <stdio.h>
#include <ctype.h>
#include <math.h>
#include <stdint.h>
#include <assert.h>
#include <dirent.h>
#include <opencv2/core.hpp>
#include <opencv2/videoio.hpp>
#include <opencv2/highgui.hpp>
#include <opencv2/imgproc.hpp>
#include <tensorflow/lite/kernels/register.h>
#include <tensorflow/lite/model.h>
#include <tensorflow/lite/interpreter.h>
#include <tensorflow/lite/delegates/hexagon/hexagon_delegate.h>
#include <tensorflow/lite/kernels/register.h>
#include <tensorflow/lite/optional_debug_tools.h>
#include <tensorflow/lite/profiling/profiler.h>
#include <tensorflow/lite/string_util.h>
#include <tensorflow/lite/tools/command_line_flags.h>
#include <tensorflow/lite/tools/delegates/delegate_provider.h>
#include <tensorflow/lite/builtin_op_data.h>
using namespace cv;
using std::cout; using std::cerr; using std::endl;
#define IMG_HEIGHT 300
#define IMG_WIDTH 300
#define IMG_DEPTH 3
#define IMG_ELEMENT_SIZE 1
#define BUFFER_SIZE (IMG_ELEMENT_SIZE * IMG_WIDTH * IMG_HEIGHT * IMG_DEPTH)
#define MODEL_NAME "/home/linaro/apps/coco_ssd_mobilenet_v1_1.0_quant_2018_06_29.tflite"
#define LABELS_FILE "/home/linaro/apps/labelmap.txt"
std::vector<std::string> labels;
int read_labels_from_file(void)
{
std::ifstream ifs(LABELS_FILE);
if (!ifs.is_open()) {
cout << "File " << LABELS_FILE << " not found" << endl;
return -1;
}
std::string line;
while (std::getline(ifs, line))
labels.push_back(line);
return 0;
}
/* Max detection Items */
#define MAX_ITEMS 10
/* Minimum Accuracy required in percentage */
#define MIN_ACCURACY 40
struct SSDbox {
float top;
float left;
float bottom;
float right;
uint32_t id;
float score;
};
static struct SSDbox SSDBoxes[MAX_ITEMS];
void SSDshowOutput(float *scores, float *mclasses, float *boxes, Mat *frame)
{
int fontface = cv::FONT_HERSHEY_PLAIN;
double scale = 1.0;
int thickness = 1;
char label[256];
int baseline = 0;
int i;
for ( i = 0; i < MAX_ITEMS; i++) {
SSDBoxes[i].top = frame->rows * boxes[i * 4];
SSDBoxes[i].left = frame->cols * boxes[i * 4 + 1];
SSDBoxes[i].bottom = frame->rows * boxes[i * 4 + 2];
SSDBoxes[i].right = frame->cols * boxes[i * 4 + 3];
SSDBoxes[i].id = (uint32_t)mclasses[i] + 1;
SSDBoxes[i].score = scores[i] * 100.00;
}
for ( i = 0; i < MAX_ITEMS; i++) {
if (SSDBoxes[i].score < MIN_ACCURACY)
break;
#if DEBUG
std::cout << i <<":\t" << SSDBoxes[i].score << ":\t" <<
labels[(SSDBoxes[i].id) ] <<":\t(" <<
(uint32_t) SSDBoxes[i].top << ", " <<
(uint32_t)SSDBoxes[i].left << ", " <<
(uint32_t) SSDBoxes[i].bottom << ", " <<
(uint32_t) SSDBoxes[i].right << ")"<< std::endl;
#endif
sprintf(label, "%02.2f%%, %s", SSDBoxes[i].score,
labels[SSDBoxes[i].id].c_str());
cv::rectangle(*frame, cv::Point((uint32_t) SSDBoxes[i].left, (uint32_t)SSDBoxes[i].top),
cv::Point( (uint32_t) SSDBoxes[i].right, (uint32_t) SSDBoxes[i].bottom),
cv::Scalar(0,255, 0), 1);
cv::Size text = cv::getTextSize(label, fontface, scale, thickness, &baseline);
cv::rectangle(*frame, cv::Point((uint32_t) SSDBoxes[i].left, (uint32_t)SSDBoxes[i].top) +
cv::Point(0, baseline), cv::Point((uint32_t) SSDBoxes[i].left, (uint32_t)SSDBoxes[i].top) +
cv::Point(text.width, -text.height), CV_RGB(0,0,0), cv::FILLED);
cv::putText(*frame, label, cv::Point((uint32_t) SSDBoxes[i].left, (uint32_t)SSDBoxes[i].top),
cv::FONT_HERSHEY_PLAIN, 1.0, cv::Scalar(57,255,20), 0.1, LINE_8);
}
}
int main(int, char**)
{
std::unique_ptr<tflite::FlatBufferModel> model;
tflite::ops::builtin::BuiltinOpResolver resolver;
std::unique_ptr<tflite::Interpreter> interpreter;
TfLiteHexagonDelegateOptions params = {0};
const char* path = "/dsp/cdsp/";
Mat frame, frame1;
size_t label_count;
char img_path[1024];
int ret, idx = 0;
char fname[16];
DIR *d;
model = tflite::FlatBufferModel::BuildFromFile(MODEL_NAME);
if (!model) {
std::cout << "Failed to mmap model ";
exit(-1);
}
tflite::InterpreterBuilder(*model, resolver)(&interpreter);
if (!interpreter) {
std::cout << "Failed to construct interpreter";
exit(-1);
}
interpreter->SetAllowFp16PrecisionForFp32(false);
interpreter->SetNumThreads(4);
std::cout << "Enabling Hexagon Delegate!\n";
TfLiteHexagonInitWithPath(path);
auto *delegate = TfLiteHexagonDelegateCreate(¶ms);
interpreter->ModifyGraphWithDelegate(delegate);
if (interpreter->AllocateTensors() != kTfLiteOk) {
std::cout << "Failed to allocate tensors!" << endl;
exit(-1);
}
const std::vector<int> inputs = interpreter->inputs();
const std::vector<int> outputs = interpreter->outputs();
int input = interpreter->inputs()[0];
/* We ONLY support UInt8 on this quantized model */
uint8_t *finput = interpreter->typed_tensor<uint8_t>(input);
float *locations, *classes, *scores, *dcount;
locations = interpreter->typed_output_tensor<float>(0);
classes = interpreter->typed_output_tensor<float>(1);
scores = interpreter->typed_output_tensor<float>(2);
dcount = interpreter->typed_output_tensor<float>(3);
if (read_labels_from_file())
exit(-1);
Mat rgb_mat(IMG_HEIGHT, IMG_WIDTH, CV_8UC3, finput);
VideoCapture capture(0);
if (!capture.isOpened())
{
cerr << "ERROR: Can't initialize camera capture" << endl;
return 1;
}
for (;;)
{
capture >> frame;
if (frame.empty())
{
cerr << "ERROR: Can't grab camera frame." << endl;
break;
}
/* resize to 300 x 300 */
cv::resize(frame, frame1, cv::Size(IMG_HEIGHT, IMG_WIDTH));
cv::cvtColor(frame1, rgb_mat, cv::COLOR_BGR2RGB);
if (interpreter->Invoke() != kTfLiteOk) {
std::cout << "Failed to invoke tflite!";
exit(-1);
}
SSDshowOutput(scores, classes, locations, &frame);
imshow("MoblieNetSSD-Demo", frame);
waitKey(1);
}
return 0;
}
|
