-
Notifications
You must be signed in to change notification settings - Fork 867
/
dnn.cpp
349 lines (273 loc) · 11.2 KB
/
dnn.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
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
#include "dnn.h"
Net Net_ReadNet(const char* model, const char* config) {
Net n = new cv::dnn::Net(cv::dnn::readNet(model, config));
return n;
}
Net Net_ReadNetBytes(const char* framework, struct ByteArray model, struct ByteArray config) {
std::vector<uchar> modelv(model.data, model.data + model.length);
std::vector<uchar> configv(config.data, config.data + config.length);
Net n = new cv::dnn::Net(cv::dnn::readNet(framework, modelv, configv));
return n;
}
Net Net_ReadNetFromCaffe(const char* prototxt, const char* caffeModel) {
Net n = new cv::dnn::Net(cv::dnn::readNetFromCaffe(prototxt, caffeModel));
return n;
}
Net Net_ReadNetFromCaffeBytes(struct ByteArray prototxt, struct ByteArray caffeModel) {
Net n = new cv::dnn::Net(cv::dnn::readNetFromCaffe(prototxt.data, prototxt.length,
caffeModel.data, caffeModel.length));
return n;
}
Net Net_ReadNetFromTensorflow(const char* model) {
Net n = new cv::dnn::Net(cv::dnn::readNetFromTensorflow(model));
return n;
}
Net Net_ReadNetFromTensorflowBytes(struct ByteArray model) {
Net n = new cv::dnn::Net(cv::dnn::readNetFromTensorflow(model.data, model.length));
return n;
}
Net Net_ReadNetFromTorch(const char* model) {
Net n = new cv::dnn::Net(cv::dnn::readNetFromTorch(model));
return n;
}
Net Net_ReadNetFromONNX(const char* model) {
Net n = new cv::dnn::Net(cv::dnn::readNetFromONNX(model));
return n;
}
Net Net_ReadNetFromONNXBytes(struct ByteArray model) {
Net n = new cv::dnn::Net(cv::dnn::readNetFromONNX(model.data, model.length));
return n;
}
void Net_Close(Net net) {
delete net;
}
bool Net_Empty(Net net) {
return net->empty();
}
void Net_SetInput(Net net, Mat blob, const char* name) {
net->setInput(*blob, name);
}
Mat Net_Forward(Net net, const char* outputName) {
return new cv::Mat(net->forward(outputName));
}
void Net_ForwardLayers(Net net, struct Mats* outputBlobs, struct CStrings outBlobNames) {
std::vector< cv::Mat > blobs;
std::vector< cv::String > names;
for (int i = 0; i < outBlobNames.length; ++i) {
names.push_back(cv::String(outBlobNames.strs[i]));
}
net->forward(blobs, names);
// copy blobs into outputBlobs
outputBlobs->mats = new Mat[blobs.size()];
for (size_t i = 0; i < blobs.size(); ++i) {
outputBlobs->mats[i] = new cv::Mat(blobs[i]);
}
outputBlobs->length = (int)blobs.size();
}
void Net_SetPreferableBackend(Net net, int backend) {
net->setPreferableBackend(backend);
}
void Net_SetPreferableTarget(Net net, int target) {
net->setPreferableTarget(target);
}
int64_t Net_GetPerfProfile(Net net) {
std::vector<double> layersTimes;
return net->getPerfProfile(layersTimes);
}
void Net_GetUnconnectedOutLayers(Net net, IntVector* res) {
std::vector< int > cids(net->getUnconnectedOutLayers());
int* ids = new int[cids.size()];
for (size_t i = 0; i < cids.size(); ++i) {
ids[i] = cids[i];
}
res->length = cids.size();
res->val = ids;
return;
}
void Net_GetLayerNames(Net net, CStrings* names) {
std::vector< cv::String > cstrs(net->getLayerNames());
const char **strs = new const char*[cstrs.size()];
for (size_t i = 0; i < cstrs.size(); ++i) {
strs[i] = cstrs[i].c_str();
}
names->length = cstrs.size();
names->strs = strs;
return;
}
struct Rect Net_BlobRectToImageRect(struct Rect rect, Size originalSize, double scalefactor, Size size, Scalar mean, bool swapRB,
int ddepth, int dataLayout, int paddingMode, Scalar borderValue) {
cv::Scalar sf(scalefactor);
cv::Size sz(size.width, size.height);
cv::Scalar cm(mean.val1, mean.val2, mean.val3, mean.val4);
cv::dnn::DataLayout dl = static_cast<cv::dnn::DataLayout>(dataLayout);
cv::dnn::ImagePaddingMode pm = static_cast<cv::dnn::ImagePaddingMode>(paddingMode);
cv::Scalar bv(borderValue.val1, borderValue.val2, borderValue.val3, borderValue.val4);
cv::dnn::Image2BlobParams params = cv::dnn::Image2BlobParams(sf, sz, cm, swapRB, ddepth, dl, pm, bv);
cv::Rect bRect = params.blobRectToImageRect(cv::Rect(rect.x, rect.y, rect.width, rect.height), cv::Size(originalSize.width, originalSize.height));
Rect r = {bRect.x, bRect.y, bRect.width, bRect.height};
return r;
}
struct Rects Net_BlobRectsToImageRects(struct Rects rects, Size originalSize, double scalefactor, Size size, Scalar mean, bool swapRB,
int ddepth, int dataLayout, int paddingMode, Scalar borderValue) {
std::vector<cv::Rect> _cRects;
for (int i = 0; i < rects.length; ++i) {
_cRects.push_back(cv::Rect(
rects.rects[i].x,
rects.rects[i].y,
rects.rects[i].width,
rects.rects[i].height
));
}
cv::Scalar sf(scalefactor);
cv::Size sz(size.width, size.height);
cv::Scalar cm(mean.val1, mean.val2, mean.val3, mean.val4);
cv::dnn::DataLayout dl = static_cast<cv::dnn::DataLayout>(dataLayout);
cv::dnn::ImagePaddingMode pm = static_cast<cv::dnn::ImagePaddingMode>(paddingMode);
cv::Scalar bv(borderValue.val1, borderValue.val2, borderValue.val3, borderValue.val4);
cv::dnn::Image2BlobParams params = cv::dnn::Image2BlobParams(sf, sz, cm, swapRB, ddepth, dl, pm, bv);
std::vector<cv::Rect> detected;
params.blobRectsToImageRects(_cRects, detected, cv::Size(originalSize.width, originalSize.height));
Rect* drects = new Rect[detected.size()];
for (size_t i = 0; i < detected.size(); ++i) {
Rect r = {detected[i].x, detected[i].y, detected[i].width, detected[i].height};
drects[i] = r;
}
Rects ret = {drects, (int)detected.size()};
return ret;
}
Mat Net_BlobFromImage(Mat image, double scalefactor, Size size, Scalar mean, bool swapRB,
bool crop) {
cv::Size sz(size.width, size.height);
cv::Scalar cm(mean.val1, mean.val2, mean.val3, mean.val4);
// use the default target ddepth here.
return new cv::Mat(cv::dnn::blobFromImage(*image, scalefactor, sz, cm, swapRB, crop));
}
Mat Net_BlobFromImageWithParams(Mat image, double scalefactor, Size size, Scalar mean, bool swapRB,
int ddepth, int dataLayout, int paddingMode, Scalar borderValue) {
cv::Scalar sf(scalefactor);
cv::Size sz(size.width, size.height);
cv::Scalar cm(mean.val1, mean.val2, mean.val3, mean.val4);
cv::dnn::DataLayout dl = static_cast<cv::dnn::DataLayout>(dataLayout);
cv::dnn::ImagePaddingMode pm = static_cast<cv::dnn::ImagePaddingMode>(paddingMode);
cv::Scalar bv(borderValue.val1, borderValue.val2, borderValue.val3, borderValue.val4);
cv::dnn::Image2BlobParams params = cv::dnn::Image2BlobParams(sf, sz, cm, swapRB, ddepth, dl, pm, bv);
return new cv::Mat(cv::dnn::blobFromImageWithParams(*image, params));
}
void Net_BlobFromImages(struct Mats images, Mat blob, double scalefactor, Size size,
Scalar mean, bool swapRB, bool crop, int ddepth) {
std::vector<cv::Mat> imgs;
for (int i = 0; i < images.length; ++i) {
imgs.push_back(*images.mats[i]);
}
cv::Size sz(size.width, size.height);
cv::Scalar cm = cv::Scalar(mean.val1, mean.val2, mean.val3, mean.val4);
// ignore the passed in ddepth, just use default.
cv::dnn::blobFromImages(imgs, *blob, scalefactor, sz, cm, swapRB, crop);
}
void Net_BlobFromImagesWithParams(struct Mats images, Mat blob, double scalefactor, Size size,
Scalar mean, bool swapRB, int ddepth, int dataLayout, int paddingMode, Scalar borderValue) {
std::vector<cv::Mat> imgs;
for (int i = 0; i < images.length; ++i) {
imgs.push_back(*images.mats[i]);
}
cv::Scalar sf(scalefactor);
cv::Size sz(size.width, size.height);
cv::Scalar cm(mean.val1, mean.val2, mean.val3, mean.val4);
cv::dnn::DataLayout dl = static_cast<cv::dnn::DataLayout>(dataLayout);
cv::dnn::ImagePaddingMode pm = static_cast<cv::dnn::ImagePaddingMode>(paddingMode);
cv::Scalar bv(borderValue.val1, borderValue.val2, borderValue.val3, borderValue.val4);
cv::dnn::Image2BlobParams params = cv::dnn::Image2BlobParams(sf, sz, cm, swapRB, ddepth, dl, pm, bv);
cv::dnn::blobFromImagesWithParams(imgs, *blob, params);
}
void Net_ImagesFromBlob(Mat blob_, struct Mats* images_) {
std::vector<cv::Mat> imgs;
cv::dnn::imagesFromBlob(*blob_, imgs);
images_->mats = new Mat[imgs.size()];
for (size_t i = 0; i < imgs.size(); ++i) {
images_->mats[i] = new cv::Mat(imgs[i]);
}
images_->length = (int) imgs.size();
}
Mat Net_GetBlobChannel(Mat blob, int imgidx, int chnidx) {
size_t w = blob->size[3];
size_t h = blob->size[2];
return new cv::Mat(h, w, CV_32F, blob->ptr<float>(imgidx, chnidx));
}
Scalar Net_GetBlobSize(Mat blob) {
Scalar scal = Scalar();
scal.val1 = blob->size[0];
scal.val2 = blob->size[1];
scal.val3 = blob->size[2];
scal.val4 = blob->size[3];
return scal;
}
Layer Net_GetLayer(Net net, int layerid) {
return new cv::Ptr<cv::dnn::Layer>(net->getLayer(layerid));
}
void Layer_Close(Layer layer) {
delete layer;
}
int Layer_InputNameToIndex(Layer layer, const char* name) {
return (*layer)->inputNameToIndex(name);
}
int Layer_OutputNameToIndex(Layer layer, const char* name) {
return (*layer)->outputNameToIndex(name);
}
const char* Layer_GetName(Layer layer) {
return (*layer)->name.c_str();
}
const char* Layer_GetType(Layer layer) {
return (*layer)->type.c_str();
}
void NMSBoxes(struct Rects bboxes, FloatVector scores, float score_threshold, float nms_threshold, IntVector* indices) {
std::vector<cv::Rect> _bboxes;
for (int i = 0; i < bboxes.length; ++i) {
_bboxes.push_back(cv::Rect(
bboxes.rects[i].x,
bboxes.rects[i].y,
bboxes.rects[i].width,
bboxes.rects[i].height
));
}
std::vector<float> _scores;
float* f;
int i;
for (i = 0, f = scores.val; i < scores.length; ++f, ++i) {
_scores.push_back(*f);
}
std::vector<int> _indices(indices->length);
cv::dnn::NMSBoxes(_bboxes, _scores, score_threshold, nms_threshold, _indices, 1.f, 0);
int* ptr = new int[_indices.size()];
for (size_t i=0; i<_indices.size(); ++i) {
ptr[i] = _indices[i];
}
indices->length = _indices.size();
indices->val = ptr;
return;
}
void NMSBoxesWithParams(struct Rects bboxes, FloatVector scores, const float score_threshold, const float nms_threshold, IntVector* indices, const float eta, const int top_k) {
std::vector<cv::Rect> _bboxes;
for (int i = 0; i < bboxes.length; ++i) {
_bboxes.push_back(cv::Rect(
bboxes.rects[i].x,
bboxes.rects[i].y,
bboxes.rects[i].width,
bboxes.rects[i].height
));
}
std::vector<float> _scores;
float* f;
int i;
for (i = 0, f = scores.val; i < scores.length; ++f, ++i) {
_scores.push_back(*f);
}
std::vector<int> _indices(indices->length);
cv::dnn::NMSBoxes(_bboxes, _scores, score_threshold, nms_threshold, _indices, eta, top_k);
int* ptr = new int[_indices.size()];
for (size_t i=0; i<_indices.size(); ++i) {
ptr[i] = _indices[i];
}
indices->length = _indices.size();
indices->val = ptr;
return;
}