-
Notifications
You must be signed in to change notification settings - Fork 0
/
depthMap_pipelined.cpp
322 lines (274 loc) · 8.99 KB
/
depthMap_pipelined.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
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
/*
* Copyright 2018 Digital Media Professionals Inc.
*
* Licensed under the Apache License, Version 2.0 (the "License");
* you may not use this file except in compliance with the License.
* You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*/
#include <fcntl.h>
#include <stdlib.h>
#include <string.h>
#include <sys/ioctl.h>
#include <sys/mman.h>
#include <sys/time.h>
#include <sys/types.h>
#include <unistd.h>
#include <algorithm>
#include <cmath>
#include <fstream>
#include <iomanip>
#include <iostream>
#include <limits>
#include <sstream>
#include <string>
#include <vector>
#include <opencv2/core/core.hpp>
#include <opencv2/imgproc/imgproc.hpp>
#include <opencv2/highgui/highgui.hpp>
#include <opencv2/imgcodecs/imgcodecs.hpp>
#include <ctime>
#include "KerasDepthMap_gen.h"
#include "util_draw.h"
#include "util_input.h"
#include "demo_common.h"
#define DMP_MEASURE_TIME
#include "util_time.h"
using namespace std;
using namespace dmp;
using namespace util;
#define SCREEN_W (get_screen_width())
#define SCREEN_H (get_screen_height())
#define IMAGE_W 768
#define IMAGE_H 256
#define IMAGE_RZ_W 384
#define IMAGE_RZ_H 128
#define FILENAME_WEIGHTS "../KerasDepthMap_weights.bin"
#define RING_BUF_SIZE 5
#define USLEEP_TIME 100000
// Define CNN network model object
CKerasDepthMap network;
int exit_code = -1;
bool do_pause = false;
// Buffer for pre-processed image data
__fp16 imgProc[RING_BUF_SIZE][IMAGE_RZ_W * IMAGE_RZ_H * 3];
// Buffer for overlay_input
COverlayRGB *overlay_input[RING_BUF_SIZE];
// Buffer for network output
vector<float> net_output[RING_BUF_SIZE];
unsigned preproc_rbuf_idx = 0;
unsigned inreference_rbuf_idx = 0;
unsigned postproc_rbuf_idx = 0;
void increment_circular_variable(unsigned &v, unsigned max) {
v++;
if (v > max) {
v = 0;
}
}
/**
* @brief Convert opencv image format to dmp board frame format
*
* @param input_frm opencv image format
* @param output_frm dmp board image format
* @param isColor input_fram is color or grayscale
*/
void opencv2dmp(cv::Mat& input_frm, COverlayRGB& output_frm, bool isColor = true) {
if (input_frm.cols != IMAGE_RZ_W && input_frm.rows != IMAGE_RZ_H){
output_frm.alloc_mem_overlay(input_frm.cols, input_frm.rows);
}
// int i = 0;
for (unsigned int h=0;h<output_frm.get_overlay_height();h++) {
for(unsigned int w=0;w<output_frm.get_overlay_width();w++) {
if (isColor){
cv::Vec3b intensity = input_frm.at<cv::Vec3b>(h, w);
uchar blue = intensity.val[0];
uchar green = intensity.val[1];
uchar red = intensity.val[2];
output_frm.set_pixel(w,h, red, green, blue);
// if (++i<32)
// printf("%d %d %d\n", red, green, blue);
}else{
uchar intensity = input_frm.at<uchar>(h, w);
output_frm.set_pixel(w,h, intensity, intensity, intensity);
}
}
}
}
void *preproc(void *) {
uint32_t imgView[IMAGE_W * IMAGE_H];
unsigned &rbuf_idx = preproc_rbuf_idx;
// Get input images filenames
vector<string> image_names;
get_jpeg_image_names("./images/", image_names);
int num_images = image_names.size();
if (num_images == 0) {
cout << "No input images." << endl;
exit_code = 2;
return NULL;
}
unsigned image_nr = 0;
while (exit_code == -1) {
while ((rbuf_idx + 1) % RING_BUF_SIZE == postproc_rbuf_idx) {
usleep(USLEEP_TIME);
}
DECLARE_TVAL(preproc);
GET_TVAL_START(preproc);
// If not pause, decode next JPEG image and do pre-processing
if (!do_pause) {
decode_jpg_file(image_names[image_nr], imgView, IMAGE_W, IMAGE_H);
overlay_input[rbuf_idx]->convert_to_overlay_pixel_format(imgView, IMAGE_W * IMAGE_H);
preproc_image(imgView, imgProc[rbuf_idx], IMAGE_W, IMAGE_H, IMAGE_RZ_W, IMAGE_RZ_H,
0, 0, 0, 1.0 / 255.0, true);
increment_circular_variable(image_nr, num_images - 1);
}
GET_SHOW_TVAL_END(preproc);
increment_circular_variable(rbuf_idx, RING_BUF_SIZE - 1);
}
return NULL;
}
void *inference(void *) {
unsigned &rbuf_idx = inreference_rbuf_idx;
// Initialize network object
while (exit_code == -1) {
while (rbuf_idx == preproc_rbuf_idx) {
usleep(USLEEP_TIME);
}
DECLARE_TVAL(inference);
GET_TVAL_START(inference);
// Run network in HW
memcpy(network.get_network_input_addr_cpu(), imgProc[rbuf_idx], IMAGE_RZ_W * IMAGE_RZ_H * 6);
network.RunNetwork();
// Handle output from HW
network.get_final_output(net_output[rbuf_idx]);
GET_SHOW_TVAL_END(inference);
increment_circular_variable(rbuf_idx, RING_BUF_SIZE - 1);
}
return NULL;
}
void *postproc(void *) {
unsigned &rbuf_idx = postproc_rbuf_idx;
// Get HW module frequency
string conv_freq;
conv_freq = std::to_string(network.get_dv_info().conv_freq);
// Create background, image, and output overlay
COverlayRGB bg_overlay(SCREEN_W, SCREEN_H);
bg_overlay.alloc_mem_overlay(SCREEN_W, SCREEN_H);
bg_overlay.load_ppm_img("fpgatitle");
COverlayRGB overlay_output(SCREEN_W, SCREEN_H);
overlay_output.alloc_mem_overlay(IMAGE_W, IMAGE_H);
// Draw background two times for front and back buffer
const char *titles[] = {
"Depth Map",
"Per-Pixel Depth Estimation",
};
for (int i = 0; i < 2; ++i) {
bg_overlay.print_to_display(0, 0);
print_demo_title(bg_overlay, titles);
swap_buffer();
}
#ifdef DMP_MEASURE_TIME
DECLARE_TVAL(swap_buffer);
TVAL_START(swap_buffer).tv_sec = 0;
#endif
while (exit_code == -1) {
while (rbuf_idx == inreference_rbuf_idx) {
usleep(USLEEP_TIME);
}
DECLARE_TVAL(postproc);
GET_TVAL_START(postproc);
// The values returned from get_final_output() is still transposed (height first) format.
// So it is actually a width=128, height=384 image
// need to transpose the output before you can compare to the Keras output.
vector<float> networkOutput_transposed(IMAGE_RZ_H*IMAGE_RZ_W);
for(int y = 0 ; y < IMAGE_RZ_H; y++) {
for(int x = 0 ; x < IMAGE_RZ_W; x++) {
networkOutput_transposed[x+y*IMAGE_RZ_W] = net_output[rbuf_idx][y+x*IMAGE_RZ_H]*255;
}
}
// Convert depth to color map
cv::Mat matDepth(IMAGE_RZ_H, IMAGE_RZ_W, CV_32FC1, networkOutput_transposed.data());
cv::Mat matDepth_8UC1;
matDepth.convertTo(matDepth_8UC1, CV_8U);
cv::Mat matDepth_8UC3, matDepth_color;
cv::cvtColor(matDepth_8UC1,matDepth_8UC3,CV_GRAY2RGB);
cv::applyColorMap(matDepth_8UC3, matDepth_color, cv::ColormapTypes::COLORMAP_JET);
cv::Mat matDepth_color_resized;
cv::resize(matDepth_color , matDepth_color_resized , cv::Size(IMAGE_W, IMAGE_H), 0, 0, CV_INTER_LINEAR);
opencv2dmp(matDepth_color_resized, overlay_output );
// Draw results
overlay_input[rbuf_idx]->print_to_display(((SCREEN_W - IMAGE_W) / 2), ((SCREEN_H - IMAGE_H) / 2)-110);
overlay_output.print_to_display(((SCREEN_W - IMAGE_W) / 2), ((SCREEN_H + IMAGE_H) / 2)-110);
// Output HW processing times
int conv_time_tot = network.get_conv_usec();
print_conv_time(bg_overlay, 8 * SCREEN_H / 9 + 10, conv_time_tot, conv_freq);
swap_buffer();
#ifdef DMP_MEASURE_TIME
GET_TVAL_END(swap_buffer);
if (TVAL_START(swap_buffer).tv_sec) {
SHOW_TIME(swap_buffer);
}
TVAL_START(swap_buffer).tv_sec = TVAL_END(swap_buffer).tv_sec;
TVAL_START(swap_buffer).tv_usec = TVAL_END(swap_buffer).tv_usec;
cout << endl;
#endif
handle_keyboard_input(exit_code, do_pause);
GET_SHOW_TVAL_END(postproc);
increment_circular_variable(rbuf_idx, RING_BUF_SIZE - 1);
}
return NULL;
}
int main(int argc, char** argv) {
// Initialize FB
if (!init_fb()) {
cerr << "init_fb() failed." << endl;
exit_code = 1;
goto error;
}
for(int i = 0; i < RING_BUF_SIZE; i++) {
overlay_input[i] = new COverlayRGB(SCREEN_W, SCREEN_H);
if (!overlay_input[i]) {
cerr << "fail to allocate COverlayRGB" << endl;
exit_code = 1;
goto error;
}
overlay_input[i]->alloc_mem_overlay(IMAGE_W, IMAGE_H);
}
// Initialize Network
network.Verbose(0);
if (!network.Initialize()) {
exit_code = 1;
goto error;
}
if (!network.LoadWeights(FILENAME_WEIGHTS)) {
exit_code = 1;
goto error;
}
if (!network.Commit()) {
exit_code = 1;
goto error;
}
pthread_t preproc_th;
pthread_t inf_th;
pthread_t postproc_th;
pthread_create(&preproc_th, NULL, preproc, NULL);
pthread_create(&inf_th, NULL, inference, NULL);
pthread_create(&postproc_th, NULL, postproc, NULL);
pthread_join(preproc_th, NULL);
pthread_join(inf_th, NULL);
pthread_join(postproc_th, NULL);
error:
shutdown();
for(int i = 0; i < RING_BUF_SIZE; i++) {
if (overlay_input[i]) {
delete overlay_input[i];
}
}
return exit_code;
}