-
Notifications
You must be signed in to change notification settings - Fork 3k
/
segnet.cpp
313 lines (246 loc) · 8.02 KB
/
segnet.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
/*
* Copyright (c) 2020, NVIDIA CORPORATION. All rights reserved.
*
* Permission is hereby granted, free of charge, to any person obtaining a
* copy of this software and associated documentation files (the "Software"),
* to deal in the Software without restriction, including without limitation
* the rights to use, copy, modify, merge, publish, distribute, sublicense,
* and/or sell copies of the Software, and to permit persons to whom the
* Software is furnished to do so, subject to the following conditions:
*
* The above copyright notice and this permission notice shall be included in
* all copies or substantial portions of the Software.
*
* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
* IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
* FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
* THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
* LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
* FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
* DEALINGS IN THE SOFTWARE.
*/
#include "videoSource.h"
#include "videoOutput.h"
#include "cudaOverlay.h"
#include "cudaMappedMemory.h"
#include "segNet.h"
#include <signal.h>
#ifdef HEADLESS
#define IS_HEADLESS() "headless" // run without display
#define DEFAULT_VISUALIZATION "overlay" // output overlay only
#else
#define IS_HEADLESS() (const char*)NULL // use display (if attached)
#define DEFAULT_VISUALIZATION "overlay|mask" // output overlay + mask
#endif
bool signal_recieved = false;
void sig_handler(int signo)
{
if( signo == SIGINT )
{
LogVerbose("received SIGINT\n");
signal_recieved = true;
}
}
int usage()
{
printf("usage: segnet [--help] [--network NETWORK] ...\n");
printf(" input_URI [output_URI]\n\n");
printf("Segment and classify a video/image stream using a semantic segmentation DNN.\n");
printf("See below for additional arguments that may not be shown above.\n\n");
printf("positional arguments:\n");
printf(" input_URI resource URI of input stream (see videoSource below)\n");
printf(" output_URI resource URI of output stream (see videoOutput below)\n\n");
printf("%s\n", segNet::Usage());
printf("%s\n", videoSource::Usage());
printf("%s\n", videoOutput::Usage());
printf("%s\n", Log::Usage());
return 0;
}
//
// segmentation buffers
//
typedef uchar3 pixelType; // this can be uchar3, uchar4, float3, float4
pixelType* imgMask = NULL; // color of each segmentation class
pixelType* imgOverlay = NULL; // input + alpha-blended mask
pixelType* imgComposite = NULL; // overlay with mask next to it
pixelType* imgOutput = NULL; // reference to one of the above three
int2 maskSize;
int2 overlaySize;
int2 compositeSize;
int2 outputSize;
// allocate mask/overlay output buffers
bool allocBuffers( int width, int height, uint32_t flags )
{
// check if the buffers were already allocated for this size
if( imgOverlay != NULL && width == overlaySize.x && height == overlaySize.y )
return true;
// free previous buffers if they exit
CUDA_FREE_HOST(imgMask);
CUDA_FREE_HOST(imgOverlay);
CUDA_FREE_HOST(imgComposite);
// allocate overlay image
overlaySize = make_int2(width, height);
if( flags & segNet::VISUALIZE_OVERLAY )
{
if( !cudaAllocMapped(&imgOverlay, overlaySize) )
{
LogError("segnet: failed to allocate CUDA memory for overlay image (%ux%u)\n", width, height);
return false;
}
imgOutput = imgOverlay;
outputSize = overlaySize;
}
// allocate mask image (half the size, unless it's the only output)
if( flags & segNet::VISUALIZE_MASK )
{
maskSize = (flags & segNet::VISUALIZE_OVERLAY) ? make_int2(width/2, height/2) : overlaySize;
if( !cudaAllocMapped(&imgMask, maskSize) )
{
LogError("segnet: failed to allocate CUDA memory for mask image\n");
return false;
}
imgOutput = imgMask;
outputSize = maskSize;
}
// allocate composite image if both overlay and mask are used
if( (flags & segNet::VISUALIZE_OVERLAY) && (flags & segNet::VISUALIZE_MASK) )
{
compositeSize = make_int2(overlaySize.x + maskSize.x, overlaySize.y);
if( !cudaAllocMapped(&imgComposite, compositeSize) )
{
LogError("segnet: failed to allocate CUDA memory for composite image\n");
return false;
}
imgOutput = imgComposite;
outputSize = compositeSize;
}
return true;
}
int main( int argc, char** argv )
{
/*
* parse command line
*/
commandLine cmdLine(argc, argv, IS_HEADLESS());
if( cmdLine.GetFlag("help") )
return usage();
/*
* attach signal handler
*/
if( signal(SIGINT, sig_handler) == SIG_ERR )
LogError("can't catch SIGINT\n");
/*
* create input stream
*/
videoSource* input = videoSource::Create(cmdLine, ARG_POSITION(0));
if( !input )
{
LogError("segnet: failed to create input stream\n");
return 1;
}
/*
* create output stream
*/
videoOutput* output = videoOutput::Create(cmdLine, ARG_POSITION(1));
if( !output )
{
LogError("segnet: failed to create output stream\n");
return 1;
}
/*
* create segmentation network
*/
segNet* net = segNet::Create(cmdLine);
if( !net )
{
LogError("segnet: failed to initialize segNet\n");
return 1;
}
// set alpha blending value for classes that don't explicitly already have an alpha
net->SetOverlayAlpha(cmdLine.GetFloat("alpha", 150.0f));
// get the desired overlay/mask filtering mode
const segNet::FilterMode filterMode = segNet::FilterModeFromStr(cmdLine.GetString("filter-mode", "linear"));
// get the visualization flags
const uint32_t visualizationFlags = segNet::VisualizationFlagsFromStr(cmdLine.GetString("visualize", DEFAULT_VISUALIZATION));
// get the object class to ignore (if any)
const char* ignoreClass = cmdLine.GetString("ignore-class", "void");
/*
* processing loop
*/
while( !signal_recieved )
{
// capture next image
pixelType* imgInput = NULL;
int status = 0;
if( !input->Capture(&imgInput, &status) )
{
if( status == videoSource::TIMEOUT )
continue;
break; // EOS
}
// allocate buffers for this size frame
if( !allocBuffers(input->GetWidth(), input->GetHeight(), visualizationFlags) )
{
LogError("segnet: failed to allocate buffers\n");
continue;
}
// process the segmentation network
if( !net->Process(imgInput, input->GetWidth(), input->GetHeight(), ignoreClass) )
{
LogError("segnet: failed to process segmentation\n");
continue;
}
// generate overlay
if( visualizationFlags & segNet::VISUALIZE_OVERLAY )
{
if( !net->Overlay(imgOverlay, overlaySize.x, overlaySize.y, filterMode) )
{
LogError("segnet: failed to process segmentation overlay.\n");
continue;
}
}
// generate mask
if( visualizationFlags & segNet::VISUALIZE_MASK )
{
if( !net->Mask(imgMask, maskSize.x, maskSize.y, filterMode) )
{
LogError("segnet:-console: failed to process segmentation mask.\n");
continue;
}
}
// generate composite
if( (visualizationFlags & segNet::VISUALIZE_OVERLAY) && (visualizationFlags & segNet::VISUALIZE_MASK) )
{
CUDA(cudaOverlay(imgOverlay, overlaySize, imgComposite, compositeSize, 0, 0));
CUDA(cudaOverlay(imgMask, maskSize, imgComposite, compositeSize, overlaySize.x, 0));
}
// render outputs
if( output != NULL )
{
output->Render(imgOutput, outputSize.x, outputSize.y);
// update the status bar
char str[256];
sprintf(str, "TensorRT %i.%i.%i | %s | Network %.0f FPS", NV_TENSORRT_MAJOR, NV_TENSORRT_MINOR, NV_TENSORRT_PATCH, net->GetNetworkName(), net->GetNetworkFPS());
output->SetStatus(str);
// check if the user quit
if( !output->IsStreaming() )
break;
}
// wait for the GPU to finish
CUDA(cudaDeviceSynchronize());
// print out timing info
net->PrintProfilerTimes();
}
/*
* destroy resources
*/
LogVerbose("segnet: shutting down...\n");
SAFE_DELETE(input);
SAFE_DELETE(output);
SAFE_DELETE(net);
CUDA_FREE_HOST(imgMask);
CUDA_FREE_HOST(imgOverlay);
CUDA_FREE_HOST(imgComposite);
LogVerbose("segnet: shutdown complete.\n");
return 0;
}