Line |
Branch |
Exec |
Source |
1 |
|
|
/* |
2 |
|
|
* Copyright (c) 2020 |
3 |
|
|
* |
4 |
|
|
* This file is part of FFmpeg. |
5 |
|
|
* |
6 |
|
|
* FFmpeg is free software; you can redistribute it and/or |
7 |
|
|
* modify it under the terms of the GNU Lesser General Public |
8 |
|
|
* License as published by the Free Software Foundation; either |
9 |
|
|
* version 2.1 of the License, or (at your option) any later version. |
10 |
|
|
* |
11 |
|
|
* FFmpeg is distributed in the hope that it will be useful, |
12 |
|
|
* but WITHOUT ANY WARRANTY; without even the implied warranty of |
13 |
|
|
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU |
14 |
|
|
* Lesser General Public License for more details. |
15 |
|
|
* |
16 |
|
|
* You should have received a copy of the GNU Lesser General Public |
17 |
|
|
* License along with FFmpeg; if not, write to the Free Software |
18 |
|
|
* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA |
19 |
|
|
*/ |
20 |
|
|
|
21 |
|
|
#include "dnn_io_proc.h" |
22 |
|
|
#include "libavutil/imgutils.h" |
23 |
|
|
#include "libswscale/swscale.h" |
24 |
|
|
#include "libavutil/avassert.h" |
25 |
|
|
#include "libavutil/detection_bbox.h" |
26 |
|
|
|
27 |
|
✗ |
static int get_datatype_size(DNNDataType dt) |
28 |
|
|
{ |
29 |
|
✗ |
switch (dt) |
30 |
|
|
{ |
31 |
|
✗ |
case DNN_FLOAT: |
32 |
|
✗ |
return sizeof(float); |
33 |
|
✗ |
case DNN_UINT8: |
34 |
|
✗ |
return sizeof(uint8_t); |
35 |
|
✗ |
default: |
36 |
|
✗ |
av_assert0(!"not supported yet."); |
37 |
|
|
return 1; |
38 |
|
|
} |
39 |
|
|
} |
40 |
|
|
|
41 |
|
✗ |
int ff_proc_from_dnn_to_frame(AVFrame *frame, DNNData *output, void *log_ctx) |
42 |
|
|
{ |
43 |
|
|
struct SwsContext *sws_ctx; |
44 |
|
✗ |
int ret = 0; |
45 |
|
✗ |
int linesize[4] = { 0 }; |
46 |
|
✗ |
void **dst_data = NULL; |
47 |
|
✗ |
void *middle_data = NULL; |
48 |
|
✗ |
uint8_t *planar_data[4] = { 0 }; |
49 |
|
✗ |
int plane_size = frame->width * frame->height * sizeof(uint8_t); |
50 |
|
✗ |
enum AVPixelFormat src_fmt = AV_PIX_FMT_NONE; |
51 |
|
✗ |
int src_datatype_size = get_datatype_size(output->dt); |
52 |
|
|
|
53 |
|
✗ |
int bytewidth = av_image_get_linesize(frame->format, frame->width, 0); |
54 |
|
✗ |
if (bytewidth < 0) { |
55 |
|
✗ |
return AVERROR(EINVAL); |
56 |
|
|
} |
57 |
|
|
/* scale == 1 and mean == 0 and dt == UINT8: passthrough */ |
58 |
|
✗ |
if (fabsf(output->scale - 1) < 1e-6f && fabsf(output->mean) < 1e-6 && output->dt == DNN_UINT8) |
59 |
|
✗ |
src_fmt = AV_PIX_FMT_GRAY8; |
60 |
|
|
/* (scale == 255 or scale == 0) and mean == 0 and dt == FLOAT: normalization */ |
61 |
|
✗ |
else if ((fabsf(output->scale - 255) < 1e-6f || fabsf(output->scale) < 1e-6f) && |
62 |
|
✗ |
fabsf(output->mean) < 1e-6 && output->dt == DNN_FLOAT) |
63 |
|
✗ |
src_fmt = AV_PIX_FMT_GRAYF32; |
64 |
|
|
else { |
65 |
|
✗ |
av_log(log_ctx, AV_LOG_ERROR, "dnn_process output data doesn't type: UINT8 " |
66 |
|
✗ |
"scale: %f, mean: %f\n", output->scale, output->mean); |
67 |
|
✗ |
return AVERROR(ENOSYS); |
68 |
|
|
} |
69 |
|
|
|
70 |
|
✗ |
dst_data = (void **)frame->data; |
71 |
|
✗ |
linesize[0] = frame->linesize[0]; |
72 |
|
✗ |
if (output->layout == DL_NCHW) { |
73 |
|
✗ |
middle_data = av_malloc(plane_size * output->dims[1]); |
74 |
|
✗ |
if (!middle_data) { |
75 |
|
✗ |
ret = AVERROR(ENOMEM); |
76 |
|
✗ |
goto err; |
77 |
|
|
} |
78 |
|
✗ |
dst_data = &middle_data; |
79 |
|
✗ |
linesize[0] = frame->width * 3; |
80 |
|
|
} |
81 |
|
|
|
82 |
|
✗ |
switch (frame->format) { |
83 |
|
✗ |
case AV_PIX_FMT_RGB24: |
84 |
|
|
case AV_PIX_FMT_BGR24: |
85 |
|
✗ |
sws_ctx = sws_getContext(frame->width * 3, |
86 |
|
|
frame->height, |
87 |
|
|
src_fmt, |
88 |
|
✗ |
frame->width * 3, |
89 |
|
|
frame->height, |
90 |
|
|
AV_PIX_FMT_GRAY8, |
91 |
|
|
0, NULL, NULL, NULL); |
92 |
|
✗ |
if (!sws_ctx) { |
93 |
|
✗ |
av_log(log_ctx, AV_LOG_ERROR, "Impossible to create scale context for the conversion " |
94 |
|
|
"fmt:%s s:%dx%d -> fmt:%s s:%dx%d\n", |
95 |
|
✗ |
av_get_pix_fmt_name(src_fmt), frame->width * 3, frame->height, |
96 |
|
✗ |
av_get_pix_fmt_name(AV_PIX_FMT_GRAY8), frame->width * 3, frame->height); |
97 |
|
✗ |
ret = AVERROR(EINVAL); |
98 |
|
✗ |
goto err; |
99 |
|
|
} |
100 |
|
✗ |
sws_scale(sws_ctx, (const uint8_t *[4]){(const uint8_t *)output->data, 0, 0, 0}, |
101 |
|
✗ |
(const int[4]){frame->width * 3 * src_datatype_size, 0, 0, 0}, 0, frame->height, |
102 |
|
|
(uint8_t * const*)dst_data, linesize); |
103 |
|
✗ |
sws_freeContext(sws_ctx); |
104 |
|
|
// convert data from planar to packed |
105 |
|
✗ |
if (output->layout == DL_NCHW) { |
106 |
|
✗ |
sws_ctx = sws_getContext(frame->width, |
107 |
|
|
frame->height, |
108 |
|
|
AV_PIX_FMT_GBRP, |
109 |
|
|
frame->width, |
110 |
|
|
frame->height, |
111 |
|
✗ |
frame->format, |
112 |
|
|
0, NULL, NULL, NULL); |
113 |
|
✗ |
if (!sws_ctx) { |
114 |
|
✗ |
av_log(log_ctx, AV_LOG_ERROR, "Impossible to create scale context for the conversion " |
115 |
|
|
"fmt:%s s:%dx%d -> fmt:%s s:%dx%d\n", |
116 |
|
|
av_get_pix_fmt_name(AV_PIX_FMT_GBRP), frame->width, frame->height, |
117 |
|
✗ |
av_get_pix_fmt_name(frame->format),frame->width, frame->height); |
118 |
|
✗ |
ret = AVERROR(EINVAL); |
119 |
|
✗ |
goto err; |
120 |
|
|
} |
121 |
|
✗ |
if (frame->format == AV_PIX_FMT_RGB24) { |
122 |
|
✗ |
planar_data[0] = (uint8_t *)middle_data + plane_size; |
123 |
|
✗ |
planar_data[1] = (uint8_t *)middle_data + plane_size * 2; |
124 |
|
✗ |
planar_data[2] = (uint8_t *)middle_data; |
125 |
|
✗ |
} else if (frame->format == AV_PIX_FMT_BGR24) { |
126 |
|
✗ |
planar_data[0] = (uint8_t *)middle_data + plane_size; |
127 |
|
✗ |
planar_data[1] = (uint8_t *)middle_data; |
128 |
|
✗ |
planar_data[2] = (uint8_t *)middle_data + plane_size * 2; |
129 |
|
|
} |
130 |
|
✗ |
sws_scale(sws_ctx, (const uint8_t * const *)planar_data, |
131 |
|
✗ |
(const int [4]){frame->width * sizeof(uint8_t), |
132 |
|
✗ |
frame->width * sizeof(uint8_t), |
133 |
|
✗ |
frame->width * sizeof(uint8_t), 0}, |
134 |
|
✗ |
0, frame->height, frame->data, frame->linesize); |
135 |
|
✗ |
sws_freeContext(sws_ctx); |
136 |
|
|
} |
137 |
|
✗ |
break; |
138 |
|
✗ |
case AV_PIX_FMT_GRAYF32: |
139 |
|
✗ |
av_image_copy_plane(frame->data[0], frame->linesize[0], |
140 |
|
✗ |
output->data, bytewidth, |
141 |
|
|
bytewidth, frame->height); |
142 |
|
✗ |
break; |
143 |
|
✗ |
case AV_PIX_FMT_YUV420P: |
144 |
|
|
case AV_PIX_FMT_YUV422P: |
145 |
|
|
case AV_PIX_FMT_YUV444P: |
146 |
|
|
case AV_PIX_FMT_YUV410P: |
147 |
|
|
case AV_PIX_FMT_YUV411P: |
148 |
|
|
case AV_PIX_FMT_GRAY8: |
149 |
|
|
case AV_PIX_FMT_NV12: |
150 |
|
✗ |
sws_ctx = sws_getContext(frame->width, |
151 |
|
|
frame->height, |
152 |
|
|
AV_PIX_FMT_GRAYF32, |
153 |
|
|
frame->width, |
154 |
|
|
frame->height, |
155 |
|
|
AV_PIX_FMT_GRAY8, |
156 |
|
|
0, NULL, NULL, NULL); |
157 |
|
✗ |
if (!sws_ctx) { |
158 |
|
✗ |
av_log(log_ctx, AV_LOG_ERROR, "Impossible to create scale context for the conversion " |
159 |
|
|
"fmt:%s s:%dx%d -> fmt:%s s:%dx%d\n", |
160 |
|
|
av_get_pix_fmt_name(src_fmt), frame->width, frame->height, |
161 |
|
|
av_get_pix_fmt_name(AV_PIX_FMT_GRAY8), frame->width, frame->height); |
162 |
|
✗ |
ret = AVERROR(EINVAL); |
163 |
|
✗ |
goto err; |
164 |
|
|
} |
165 |
|
✗ |
sws_scale(sws_ctx, (const uint8_t *[4]){(const uint8_t *)output->data, 0, 0, 0}, |
166 |
|
✗ |
(const int[4]){frame->width * src_datatype_size, 0, 0, 0}, 0, frame->height, |
167 |
|
✗ |
(uint8_t * const*)frame->data, frame->linesize); |
168 |
|
✗ |
sws_freeContext(sws_ctx); |
169 |
|
✗ |
break; |
170 |
|
✗ |
default: |
171 |
|
✗ |
avpriv_report_missing_feature(log_ctx, "%s", av_get_pix_fmt_name(frame->format)); |
172 |
|
✗ |
ret = AVERROR(ENOSYS); |
173 |
|
✗ |
goto err; |
174 |
|
|
} |
175 |
|
|
|
176 |
|
✗ |
err: |
177 |
|
✗ |
av_free(middle_data); |
178 |
|
✗ |
return ret; |
179 |
|
|
} |
180 |
|
|
|
181 |
|
✗ |
int ff_proc_from_frame_to_dnn(AVFrame *frame, DNNData *input, void *log_ctx) |
182 |
|
|
{ |
183 |
|
|
struct SwsContext *sws_ctx; |
184 |
|
✗ |
int ret = 0; |
185 |
|
✗ |
int linesize[4] = { 0 }; |
186 |
|
✗ |
void **src_data = NULL; |
187 |
|
✗ |
void *middle_data = NULL; |
188 |
|
✗ |
uint8_t *planar_data[4] = { 0 }; |
189 |
|
✗ |
int plane_size = frame->width * frame->height * sizeof(uint8_t); |
190 |
|
✗ |
enum AVPixelFormat dst_fmt = AV_PIX_FMT_NONE; |
191 |
|
✗ |
int dst_datatype_size = get_datatype_size(input->dt); |
192 |
|
✗ |
int bytewidth = av_image_get_linesize(frame->format, frame->width, 0); |
193 |
|
✗ |
if (bytewidth < 0) { |
194 |
|
✗ |
return AVERROR(EINVAL); |
195 |
|
|
} |
196 |
|
|
/* scale == 1 and mean == 0 and dt == UINT8: passthrough */ |
197 |
|
✗ |
if (fabsf(input->scale - 1) < 1e-6f && fabsf(input->mean) < 1e-6 && input->dt == DNN_UINT8) |
198 |
|
✗ |
dst_fmt = AV_PIX_FMT_GRAY8; |
199 |
|
|
/* (scale == 255 or scale == 0) and mean == 0 and dt == FLOAT: normalization */ |
200 |
|
✗ |
else if ((fabsf(input->scale - 255) < 1e-6f || fabsf(input->scale) < 1e-6f) && |
201 |
|
✗ |
fabsf(input->mean) < 1e-6 && input->dt == DNN_FLOAT) |
202 |
|
✗ |
dst_fmt = AV_PIX_FMT_GRAYF32; |
203 |
|
|
else { |
204 |
|
✗ |
av_log(log_ctx, AV_LOG_ERROR, "dnn_process input data doesn't support type: UINT8 " |
205 |
|
✗ |
"scale: %f, mean: %f\n", input->scale, input->mean); |
206 |
|
✗ |
return AVERROR(ENOSYS); |
207 |
|
|
} |
208 |
|
|
|
209 |
|
✗ |
src_data = (void **)frame->data; |
210 |
|
✗ |
linesize[0] = frame->linesize[0]; |
211 |
|
✗ |
if (input->layout == DL_NCHW) { |
212 |
|
✗ |
middle_data = av_malloc(plane_size * input->dims[1]); |
213 |
|
✗ |
if (!middle_data) { |
214 |
|
✗ |
ret = AVERROR(ENOMEM); |
215 |
|
✗ |
goto err; |
216 |
|
|
} |
217 |
|
✗ |
src_data = &middle_data; |
218 |
|
✗ |
linesize[0] = frame->width * 3; |
219 |
|
|
} |
220 |
|
|
|
221 |
|
✗ |
switch (frame->format) { |
222 |
|
✗ |
case AV_PIX_FMT_RGB24: |
223 |
|
|
case AV_PIX_FMT_BGR24: |
224 |
|
|
// convert data from planar to packed |
225 |
|
✗ |
if (input->layout == DL_NCHW) { |
226 |
|
✗ |
sws_ctx = sws_getContext(frame->width, |
227 |
|
|
frame->height, |
228 |
|
✗ |
frame->format, |
229 |
|
|
frame->width, |
230 |
|
|
frame->height, |
231 |
|
|
AV_PIX_FMT_GBRP, |
232 |
|
|
0, NULL, NULL, NULL); |
233 |
|
✗ |
if (!sws_ctx) { |
234 |
|
✗ |
av_log(log_ctx, AV_LOG_ERROR, "Impossible to create scale context for the conversion " |
235 |
|
|
"fmt:%s s:%dx%d -> fmt:%s s:%dx%d\n", |
236 |
|
✗ |
av_get_pix_fmt_name(frame->format), frame->width, frame->height, |
237 |
|
|
av_get_pix_fmt_name(AV_PIX_FMT_GBRP),frame->width, frame->height); |
238 |
|
✗ |
ret = AVERROR(EINVAL); |
239 |
|
✗ |
goto err; |
240 |
|
|
} |
241 |
|
✗ |
if (frame->format == AV_PIX_FMT_RGB24) { |
242 |
|
✗ |
planar_data[0] = (uint8_t *)middle_data + plane_size; |
243 |
|
✗ |
planar_data[1] = (uint8_t *)middle_data + plane_size * 2; |
244 |
|
✗ |
planar_data[2] = (uint8_t *)middle_data; |
245 |
|
✗ |
} else if (frame->format == AV_PIX_FMT_BGR24) { |
246 |
|
✗ |
planar_data[0] = (uint8_t *)middle_data + plane_size; |
247 |
|
✗ |
planar_data[1] = (uint8_t *)middle_data; |
248 |
|
✗ |
planar_data[2] = (uint8_t *)middle_data + plane_size * 2; |
249 |
|
|
} |
250 |
|
✗ |
sws_scale(sws_ctx, (const uint8_t * const *)frame->data, |
251 |
|
✗ |
frame->linesize, 0, frame->height, planar_data, |
252 |
|
✗ |
(const int [4]){frame->width * sizeof(uint8_t), |
253 |
|
✗ |
frame->width * sizeof(uint8_t), |
254 |
|
✗ |
frame->width * sizeof(uint8_t), 0}); |
255 |
|
✗ |
sws_freeContext(sws_ctx); |
256 |
|
|
} |
257 |
|
✗ |
sws_ctx = sws_getContext(frame->width * 3, |
258 |
|
|
frame->height, |
259 |
|
|
AV_PIX_FMT_GRAY8, |
260 |
|
✗ |
frame->width * 3, |
261 |
|
|
frame->height, |
262 |
|
|
dst_fmt, |
263 |
|
|
0, NULL, NULL, NULL); |
264 |
|
✗ |
if (!sws_ctx) { |
265 |
|
✗ |
av_log(log_ctx, AV_LOG_ERROR, "Impossible to create scale context for the conversion " |
266 |
|
|
"fmt:%s s:%dx%d -> fmt:%s s:%dx%d\n", |
267 |
|
✗ |
av_get_pix_fmt_name(AV_PIX_FMT_GRAY8), frame->width * 3, frame->height, |
268 |
|
✗ |
av_get_pix_fmt_name(dst_fmt),frame->width * 3, frame->height); |
269 |
|
✗ |
ret = AVERROR(EINVAL); |
270 |
|
✗ |
goto err; |
271 |
|
|
} |
272 |
|
✗ |
sws_scale(sws_ctx, (const uint8_t **)src_data, |
273 |
|
|
linesize, 0, frame->height, |
274 |
|
✗ |
(uint8_t * const [4]){input->data, 0, 0, 0}, |
275 |
|
✗ |
(const int [4]){frame->width * 3 * dst_datatype_size, 0, 0, 0}); |
276 |
|
✗ |
sws_freeContext(sws_ctx); |
277 |
|
✗ |
break; |
278 |
|
✗ |
case AV_PIX_FMT_GRAYF32: |
279 |
|
✗ |
av_image_copy_plane(input->data, bytewidth, |
280 |
|
✗ |
frame->data[0], frame->linesize[0], |
281 |
|
|
bytewidth, frame->height); |
282 |
|
✗ |
break; |
283 |
|
✗ |
case AV_PIX_FMT_YUV420P: |
284 |
|
|
case AV_PIX_FMT_YUV422P: |
285 |
|
|
case AV_PIX_FMT_YUV444P: |
286 |
|
|
case AV_PIX_FMT_YUV410P: |
287 |
|
|
case AV_PIX_FMT_YUV411P: |
288 |
|
|
case AV_PIX_FMT_GRAY8: |
289 |
|
|
case AV_PIX_FMT_NV12: |
290 |
|
✗ |
sws_ctx = sws_getContext(frame->width, |
291 |
|
|
frame->height, |
292 |
|
|
AV_PIX_FMT_GRAY8, |
293 |
|
|
frame->width, |
294 |
|
|
frame->height, |
295 |
|
|
dst_fmt, |
296 |
|
|
0, NULL, NULL, NULL); |
297 |
|
✗ |
if (!sws_ctx) { |
298 |
|
✗ |
av_log(log_ctx, AV_LOG_ERROR, "Impossible to create scale context for the conversion " |
299 |
|
|
"fmt:%s s:%dx%d -> fmt:%s s:%dx%d\n", |
300 |
|
|
av_get_pix_fmt_name(AV_PIX_FMT_GRAY8), frame->width, frame->height, |
301 |
|
|
av_get_pix_fmt_name(dst_fmt),frame->width, frame->height); |
302 |
|
✗ |
ret = AVERROR(EINVAL); |
303 |
|
✗ |
goto err; |
304 |
|
|
} |
305 |
|
✗ |
sws_scale(sws_ctx, (const uint8_t **)frame->data, |
306 |
|
✗ |
frame->linesize, 0, frame->height, |
307 |
|
✗ |
(uint8_t * const [4]){input->data, 0, 0, 0}, |
308 |
|
✗ |
(const int [4]){frame->width * dst_datatype_size, 0, 0, 0}); |
309 |
|
✗ |
sws_freeContext(sws_ctx); |
310 |
|
✗ |
break; |
311 |
|
✗ |
default: |
312 |
|
✗ |
avpriv_report_missing_feature(log_ctx, "%s", av_get_pix_fmt_name(frame->format)); |
313 |
|
✗ |
ret = AVERROR(ENOSYS); |
314 |
|
✗ |
goto err; |
315 |
|
|
} |
316 |
|
✗ |
err: |
317 |
|
✗ |
av_free(middle_data); |
318 |
|
✗ |
return ret; |
319 |
|
|
} |
320 |
|
|
|
321 |
|
✗ |
static enum AVPixelFormat get_pixel_format(DNNData *data) |
322 |
|
|
{ |
323 |
|
✗ |
if (data->dt == DNN_UINT8) { |
324 |
|
✗ |
switch (data->order) { |
325 |
|
✗ |
case DCO_BGR: |
326 |
|
✗ |
return AV_PIX_FMT_BGR24; |
327 |
|
✗ |
case DCO_RGB: |
328 |
|
✗ |
return AV_PIX_FMT_RGB24; |
329 |
|
✗ |
default: |
330 |
|
✗ |
av_assert0(!"unsupported data pixel format.\n"); |
331 |
|
|
return AV_PIX_FMT_BGR24; |
332 |
|
|
} |
333 |
|
|
} |
334 |
|
|
|
335 |
|
✗ |
av_assert0(!"unsupported data type.\n"); |
336 |
|
|
return AV_PIX_FMT_BGR24; |
337 |
|
|
} |
338 |
|
|
|
339 |
|
✗ |
int ff_frame_to_dnn_classify(AVFrame *frame, DNNData *input, uint32_t bbox_index, void *log_ctx) |
340 |
|
|
{ |
341 |
|
|
const AVPixFmtDescriptor *desc; |
342 |
|
|
int offsetx[4], offsety[4]; |
343 |
|
|
uint8_t *bbox_data[4]; |
344 |
|
|
struct SwsContext *sws_ctx; |
345 |
|
|
int linesizes[4]; |
346 |
|
✗ |
int ret = 0; |
347 |
|
|
enum AVPixelFormat fmt; |
348 |
|
|
int left, top, width, height; |
349 |
|
|
int width_idx, height_idx; |
350 |
|
|
const AVDetectionBBoxHeader *header; |
351 |
|
|
const AVDetectionBBox *bbox; |
352 |
|
✗ |
AVFrameSideData *sd = av_frame_get_side_data(frame, AV_FRAME_DATA_DETECTION_BBOXES); |
353 |
|
✗ |
av_assert0(sd); |
354 |
|
|
|
355 |
|
|
/* (scale != 1 and scale != 0) or mean != 0 */ |
356 |
|
✗ |
if ((fabsf(input->scale - 1) > 1e-6f && fabsf(input->scale) > 1e-6f) || |
357 |
|
✗ |
fabsf(input->mean) > 1e-6f) { |
358 |
|
✗ |
av_log(log_ctx, AV_LOG_ERROR, "dnn_classify input data doesn't support " |
359 |
|
✗ |
"scale: %f, mean: %f\n", input->scale, input->mean); |
360 |
|
✗ |
return AVERROR(ENOSYS); |
361 |
|
|
} |
362 |
|
|
|
363 |
|
✗ |
if (input->layout == DL_NCHW) { |
364 |
|
✗ |
av_log(log_ctx, AV_LOG_ERROR, "dnn_classify input data doesn't support layout: NCHW\n"); |
365 |
|
✗ |
return AVERROR(ENOSYS); |
366 |
|
|
} |
367 |
|
|
|
368 |
|
✗ |
width_idx = dnn_get_width_idx_by_layout(input->layout); |
369 |
|
✗ |
height_idx = dnn_get_height_idx_by_layout(input->layout); |
370 |
|
|
|
371 |
|
✗ |
header = (const AVDetectionBBoxHeader *)sd->data; |
372 |
|
✗ |
bbox = av_get_detection_bbox(header, bbox_index); |
373 |
|
|
|
374 |
|
✗ |
left = bbox->x; |
375 |
|
✗ |
width = bbox->w; |
376 |
|
✗ |
top = bbox->y; |
377 |
|
✗ |
height = bbox->h; |
378 |
|
|
|
379 |
|
✗ |
fmt = get_pixel_format(input); |
380 |
|
✗ |
sws_ctx = sws_getContext(width, height, frame->format, |
381 |
|
|
input->dims[width_idx], |
382 |
|
|
input->dims[height_idx], fmt, |
383 |
|
|
SWS_FAST_BILINEAR, NULL, NULL, NULL); |
384 |
|
✗ |
if (!sws_ctx) { |
385 |
|
✗ |
av_log(log_ctx, AV_LOG_ERROR, "Failed to create scale context for the conversion " |
386 |
|
|
"fmt:%s s:%dx%d -> fmt:%s s:%dx%d\n", |
387 |
|
✗ |
av_get_pix_fmt_name(frame->format), width, height, |
388 |
|
|
av_get_pix_fmt_name(fmt), |
389 |
|
|
input->dims[width_idx], |
390 |
|
|
input->dims[height_idx]); |
391 |
|
✗ |
return AVERROR(EINVAL); |
392 |
|
|
} |
393 |
|
|
|
394 |
|
✗ |
ret = av_image_fill_linesizes(linesizes, fmt, input->dims[width_idx]); |
395 |
|
✗ |
if (ret < 0) { |
396 |
|
✗ |
av_log(log_ctx, AV_LOG_ERROR, "unable to get linesizes with av_image_fill_linesizes"); |
397 |
|
✗ |
sws_freeContext(sws_ctx); |
398 |
|
✗ |
return ret; |
399 |
|
|
} |
400 |
|
|
|
401 |
|
✗ |
desc = av_pix_fmt_desc_get(frame->format); |
402 |
|
✗ |
offsetx[1] = offsetx[2] = AV_CEIL_RSHIFT(left, desc->log2_chroma_w); |
403 |
|
✗ |
offsetx[0] = offsetx[3] = left; |
404 |
|
|
|
405 |
|
✗ |
offsety[1] = offsety[2] = AV_CEIL_RSHIFT(top, desc->log2_chroma_h); |
406 |
|
✗ |
offsety[0] = offsety[3] = top; |
407 |
|
|
|
408 |
|
✗ |
for (int k = 0; frame->data[k]; k++) |
409 |
|
✗ |
bbox_data[k] = frame->data[k] + offsety[k] * frame->linesize[k] + offsetx[k]; |
410 |
|
|
|
411 |
|
✗ |
sws_scale(sws_ctx, (const uint8_t *const *)&bbox_data, frame->linesize, |
412 |
|
|
0, height, |
413 |
|
✗ |
(uint8_t *const [4]){input->data, 0, 0, 0}, linesizes); |
414 |
|
|
|
415 |
|
✗ |
sws_freeContext(sws_ctx); |
416 |
|
|
|
417 |
|
✗ |
return ret; |
418 |
|
|
} |
419 |
|
|
|
420 |
|
✗ |
int ff_frame_to_dnn_detect(AVFrame *frame, DNNData *input, void *log_ctx) |
421 |
|
|
{ |
422 |
|
|
struct SwsContext *sws_ctx; |
423 |
|
|
int linesizes[4]; |
424 |
|
✗ |
int ret = 0, width_idx, height_idx; |
425 |
|
✗ |
enum AVPixelFormat fmt = get_pixel_format(input); |
426 |
|
|
|
427 |
|
|
/* (scale != 1 and scale != 0) or mean != 0 */ |
428 |
|
✗ |
if ((fabsf(input->scale - 1) > 1e-6f && fabsf(input->scale) > 1e-6f) || |
429 |
|
✗ |
fabsf(input->mean) > 1e-6f) { |
430 |
|
✗ |
av_log(log_ctx, AV_LOG_ERROR, "dnn_detect input data doesn't support " |
431 |
|
✗ |
"scale: %f, mean: %f\n", input->scale, input->mean); |
432 |
|
✗ |
return AVERROR(ENOSYS); |
433 |
|
|
} |
434 |
|
|
|
435 |
|
✗ |
if (input->layout == DL_NCHW) { |
436 |
|
✗ |
av_log(log_ctx, AV_LOG_ERROR, "dnn_detect input data doesn't support layout: NCHW\n"); |
437 |
|
✗ |
return AVERROR(ENOSYS); |
438 |
|
|
} |
439 |
|
|
|
440 |
|
✗ |
width_idx = dnn_get_width_idx_by_layout(input->layout); |
441 |
|
✗ |
height_idx = dnn_get_height_idx_by_layout(input->layout); |
442 |
|
|
|
443 |
|
✗ |
sws_ctx = sws_getContext(frame->width, frame->height, frame->format, |
444 |
|
|
input->dims[width_idx], |
445 |
|
|
input->dims[height_idx], fmt, |
446 |
|
|
SWS_FAST_BILINEAR, NULL, NULL, NULL); |
447 |
|
✗ |
if (!sws_ctx) { |
448 |
|
✗ |
av_log(log_ctx, AV_LOG_ERROR, "Impossible to create scale context for the conversion " |
449 |
|
|
"fmt:%s s:%dx%d -> fmt:%s s:%dx%d\n", |
450 |
|
✗ |
av_get_pix_fmt_name(frame->format), frame->width, frame->height, |
451 |
|
|
av_get_pix_fmt_name(fmt), input->dims[width_idx], |
452 |
|
|
input->dims[height_idx]); |
453 |
|
✗ |
return AVERROR(EINVAL); |
454 |
|
|
} |
455 |
|
|
|
456 |
|
✗ |
ret = av_image_fill_linesizes(linesizes, fmt, input->dims[width_idx]); |
457 |
|
✗ |
if (ret < 0) { |
458 |
|
✗ |
av_log(log_ctx, AV_LOG_ERROR, "unable to get linesizes with av_image_fill_linesizes"); |
459 |
|
✗ |
sws_freeContext(sws_ctx); |
460 |
|
✗ |
return ret; |
461 |
|
|
} |
462 |
|
|
|
463 |
|
✗ |
sws_scale(sws_ctx, (const uint8_t *const *)frame->data, frame->linesize, 0, frame->height, |
464 |
|
✗ |
(uint8_t *const [4]){input->data, 0, 0, 0}, linesizes); |
465 |
|
|
|
466 |
|
✗ |
sws_freeContext(sws_ctx); |
467 |
|
✗ |
return ret; |
468 |
|
|
} |
469 |
|
|
|