[FFmpeg-cvslog] dnn: extract common functions used by different filters

Guo, Yejun git at videolan.org
Thu Feb 18 04:27:09 EET 2021


ffmpeg | branch: master | Guo, Yejun <yejun.guo at intel.com> | Tue Jan 26 13:35:30 2021 +0800| [bdce6361004b3a24082e00b6c0126e2694a65b50] | committer: Guo, Yejun

dnn: extract common functions used by different filters

Signed-off-by: Guo, Yejun <yejun.guo at intel.com>

> http://git.videolan.org/gitweb.cgi/ffmpeg.git/?a=commit;h=bdce6361004b3a24082e00b6c0126e2694a65b50
---

 libavfilter/Makefile            |   1 +
 libavfilter/dnn_filter_common.c | 106 ++++++++++++++++++++++++++++++++++++++++
 libavfilter/dnn_filter_common.h |  59 ++++++++++++++++++++++
 libavfilter/vf_derain.c         |  46 ++++-------------
 libavfilter/vf_dnn_processing.c |  93 ++++++-----------------------------
 libavfilter/vf_sr.c             |  53 +++++---------------
 6 files changed, 201 insertions(+), 157 deletions(-)

diff --git a/libavfilter/Makefile b/libavfilter/Makefile
index 05bb2d2031..359ea7f903 100644
--- a/libavfilter/Makefile
+++ b/libavfilter/Makefile
@@ -27,6 +27,7 @@ OBJS-$(HAVE_THREADS)                         += pthread.o
 # subsystems
 OBJS-$(CONFIG_QSVVPP)                        += qsvvpp.o
 OBJS-$(CONFIG_SCENE_SAD)                     += scene_sad.o
+OBJS-$(CONFIG_DNN)                           += dnn_filter_common.o
 include $(SRC_PATH)/libavfilter/dnn/Makefile
 
 # audio filters
diff --git a/libavfilter/dnn_filter_common.c b/libavfilter/dnn_filter_common.c
new file mode 100644
index 0000000000..5d0d7d3b90
--- /dev/null
+++ b/libavfilter/dnn_filter_common.c
@@ -0,0 +1,106 @@
+/*
+ * This file is part of FFmpeg.
+ *
+ * FFmpeg is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU Lesser General Public
+ * License as published by the Free Software Foundation; either
+ * version 2.1 of the License, or (at your option) any later version.
+ *
+ * FFmpeg is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+ * Lesser General Public License for more details.
+ *
+ * You should have received a copy of the GNU Lesser General Public
+ * License along with FFmpeg; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
+ */
+
+#include "dnn_filter_common.h"
+
+int ff_dnn_init(DnnContext *ctx, AVFilterContext *filter_ctx)
+{
+    if (!ctx->model_filename) {
+        av_log(filter_ctx, AV_LOG_ERROR, "model file for network is not specified\n");
+        return AVERROR(EINVAL);
+    }
+    if (!ctx->model_inputname) {
+        av_log(filter_ctx, AV_LOG_ERROR, "input name of the model network is not specified\n");
+        return AVERROR(EINVAL);
+    }
+    if (!ctx->model_outputname) {
+        av_log(filter_ctx, AV_LOG_ERROR, "output name of the model network is not specified\n");
+        return AVERROR(EINVAL);
+    }
+
+    ctx->dnn_module = ff_get_dnn_module(ctx->backend_type);
+    if (!ctx->dnn_module) {
+        av_log(filter_ctx, AV_LOG_ERROR, "could not create DNN module for requested backend\n");
+        return AVERROR(ENOMEM);
+    }
+    if (!ctx->dnn_module->load_model) {
+        av_log(filter_ctx, AV_LOG_ERROR, "load_model for network is not specified\n");
+        return AVERROR(EINVAL);
+    }
+
+    ctx->model = (ctx->dnn_module->load_model)(ctx->model_filename, ctx->backend_options, filter_ctx);
+    if (!ctx->model) {
+        av_log(filter_ctx, AV_LOG_ERROR, "could not load DNN model\n");
+        return AVERROR(EINVAL);
+    }
+
+    if (!ctx->dnn_module->execute_model_async && ctx->async) {
+        ctx->async = 0;
+        av_log(filter_ctx, AV_LOG_WARNING, "this backend does not support async execution, roll back to sync.\n");
+    }
+
+#if !HAVE_PTHREAD_CANCEL
+    if (ctx->async) {
+        ctx->async = 0;
+        av_log(filter_ctx, AV_LOG_WARNING, "pthread is not supported, roll back to sync.\n");
+    }
+#endif
+
+    return 0;
+}
+
+DNNReturnType ff_dnn_get_input(DnnContext *ctx, DNNData *input)
+{
+    return ctx->model->get_input(ctx->model->model, input, ctx->model_inputname);
+}
+
+DNNReturnType ff_dnn_get_output(DnnContext *ctx, int input_width, int input_height, int *output_width, int *output_height)
+{
+    return ctx->model->get_output(ctx->model->model, ctx->model_inputname, input_width, input_height,
+                                    ctx->model_outputname, output_width, output_height);
+}
+
+DNNReturnType ff_dnn_execute_model(DnnContext *ctx, AVFrame *in_frame, AVFrame *out_frame)
+{
+    return (ctx->dnn_module->execute_model)(ctx->model, ctx->model_inputname, in_frame,
+                                            (const char **)&ctx->model_outputname, 1, out_frame);
+}
+
+DNNReturnType ff_dnn_execute_model_async(DnnContext *ctx, AVFrame *in_frame, AVFrame *out_frame)
+{
+    return (ctx->dnn_module->execute_model_async)(ctx->model, ctx->model_inputname, in_frame,
+                                                  (const char **)&ctx->model_outputname, 1, out_frame);
+}
+
+DNNAsyncStatusType ff_dnn_get_async_result(DnnContext *ctx, AVFrame **in_frame, AVFrame **out_frame)
+{
+    return (ctx->dnn_module->get_async_result)(ctx->model, in_frame, out_frame);
+}
+
+DNNReturnType ff_dnn_flush(DnnContext *ctx)
+{
+    return (ctx->dnn_module->flush)(ctx->model);
+}
+
+void ff_dnn_uninit(DnnContext *ctx)
+{
+    if (ctx->dnn_module) {
+        (ctx->dnn_module->free_model)(&ctx->model);
+        av_freep(&ctx->dnn_module);
+    }
+}
diff --git a/libavfilter/dnn_filter_common.h b/libavfilter/dnn_filter_common.h
new file mode 100644
index 0000000000..ab49a992ed
--- /dev/null
+++ b/libavfilter/dnn_filter_common.h
@@ -0,0 +1,59 @@
+/*
+ * This file is part of FFmpeg.
+ *
+ * FFmpeg is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU Lesser General Public
+ * License as published by the Free Software Foundation; either
+ * version 2.1 of the License, or (at your option) any later version.
+ *
+ * FFmpeg is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+ * Lesser General Public License for more details.
+ *
+ * You should have received a copy of the GNU Lesser General Public
+ * License along with FFmpeg; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
+ */
+
+/**
+ * @file
+ * common functions for the dnn based filters
+ */
+
+#ifndef AVFILTER_DNN_FILTER_COMMON_H
+#define AVFILTER_DNN_FILTER_COMMON_H
+
+#include "dnn_interface.h"
+
+typedef struct DnnContext {
+    char *model_filename;
+    DNNBackendType backend_type;
+    char *model_inputname;
+    char *model_outputname;
+    char *backend_options;
+    int async;
+
+    DNNModule *dnn_module;
+    DNNModel *model;
+} DnnContext;
+
+#define DNN_COMMON_OPTIONS \
+    { "model",              "path to model file",         OFFSET(model_filename),   AV_OPT_TYPE_STRING,    { .str = NULL }, 0, 0, FLAGS },\
+    { "input",              "input name of the model",    OFFSET(model_inputname),  AV_OPT_TYPE_STRING,    { .str = NULL }, 0, 0, FLAGS },\
+    { "output",             "output name of the model",   OFFSET(model_outputname), AV_OPT_TYPE_STRING,    { .str = NULL }, 0, 0, FLAGS },\
+    { "backend_configs",    "backend configs",            OFFSET(backend_options),  AV_OPT_TYPE_STRING,    { .str = NULL }, 0, 0, FLAGS },\
+    { "options",            "backend configs",            OFFSET(backend_options),  AV_OPT_TYPE_STRING,    { .str = NULL }, 0, 0, FLAGS },\
+    { "async",              "use DNN async inference",    OFFSET(async),            AV_OPT_TYPE_BOOL,      { .i64 = 1},     0, 1, FLAGS},
+
+
+int ff_dnn_init(DnnContext *ctx, AVFilterContext *filter_ctx);
+DNNReturnType ff_dnn_get_input(DnnContext *ctx, DNNData *input);
+DNNReturnType ff_dnn_get_output(DnnContext *ctx, int input_width, int input_height, int *output_width, int *output_height);
+DNNReturnType ff_dnn_execute_model(DnnContext *ctx, AVFrame *in_frame, AVFrame *out_frame);
+DNNReturnType ff_dnn_execute_model_async(DnnContext *ctx, AVFrame *in_frame, AVFrame *out_frame);
+DNNAsyncStatusType ff_dnn_get_async_result(DnnContext *ctx, AVFrame **in_frame, AVFrame **out_frame);
+DNNReturnType ff_dnn_flush(DnnContext *ctx);
+void ff_dnn_uninit(DnnContext *ctx);
+
+#endif
diff --git a/libavfilter/vf_derain.c b/libavfilter/vf_derain.c
index 77dd401263..ec9853d957 100644
--- a/libavfilter/vf_derain.c
+++ b/libavfilter/vf_derain.c
@@ -27,18 +27,14 @@
 #include "libavformat/avio.h"
 #include "libavutil/opt.h"
 #include "avfilter.h"
-#include "dnn_interface.h"
+#include "dnn_filter_common.h"
 #include "formats.h"
 #include "internal.h"
 
 typedef struct DRContext {
     const AVClass *class;
-
+    DnnContext dnnctx;
     int                filter_type;
-    char              *model_filename;
-    DNNBackendType     backend_type;
-    DNNModule         *dnn_module;
-    DNNModel          *model;
 } DRContext;
 
 #define OFFSET(x) offsetof(DRContext, x)
@@ -47,12 +43,14 @@ static const AVOption derain_options[] = {
     { "filter_type", "filter type(derain/dehaze)",  OFFSET(filter_type),    AV_OPT_TYPE_INT,    { .i64 = 0 },    0, 1, FLAGS, "type" },
     { "derain",      "derain filter flag",          0,                      AV_OPT_TYPE_CONST,  { .i64 = 0 },    0, 0, FLAGS, "type" },
     { "dehaze",      "dehaze filter flag",          0,                      AV_OPT_TYPE_CONST,  { .i64 = 1 },    0, 0, FLAGS, "type" },
-    { "dnn_backend", "DNN backend",                 OFFSET(backend_type),   AV_OPT_TYPE_INT,    { .i64 = 0 },    0, 1, FLAGS, "backend" },
+    { "dnn_backend", "DNN backend",                 OFFSET(dnnctx.backend_type),   AV_OPT_TYPE_INT,    { .i64 = 0 },    0, 1, FLAGS, "backend" },
     { "native",      "native backend flag",         0,                      AV_OPT_TYPE_CONST,  { .i64 = 0 },    0, 0, FLAGS, "backend" },
 #if (CONFIG_LIBTENSORFLOW == 1)
     { "tensorflow",  "tensorflow backend flag",     0,                      AV_OPT_TYPE_CONST,  { .i64 = 1 },    0, 0, FLAGS, "backend" },
 #endif
-    { "model",       "path to model file",          OFFSET(model_filename), AV_OPT_TYPE_STRING, { .str = NULL }, 0, 0, FLAGS },
+    { "model",       "path to model file",          OFFSET(dnnctx.model_filename),   AV_OPT_TYPE_STRING,    { .str = NULL }, 0, 0, FLAGS },
+    { "input",       "input name of the model",     OFFSET(dnnctx.model_inputname),  AV_OPT_TYPE_STRING,    { .str = "x" },  0, 0, FLAGS },
+    { "output",      "output name of the model",    OFFSET(dnnctx.model_outputname), AV_OPT_TYPE_STRING,    { .str = "y" },  0, 0, FLAGS },
     { NULL }
 };
 
@@ -77,7 +75,6 @@ static int filter_frame(AVFilterLink *inlink, AVFrame *in)
     AVFilterLink *outlink = ctx->outputs[0];
     DRContext *dr_context = ctx->priv;
     DNNReturnType dnn_result;
-    const char *model_output_name = "y";
     AVFrame *out;
 
     out = ff_get_video_buffer(outlink, outlink->w, outlink->h);
@@ -88,7 +85,7 @@ static int filter_frame(AVFilterLink *inlink, AVFrame *in)
     }
     av_frame_copy_props(out, in);
 
-    dnn_result = (dr_context->dnn_module->execute_model)(dr_context->model, "x", in, &model_output_name, 1, out);
+    dnn_result = ff_dnn_execute_model(&dr_context->dnnctx, in, out);
     if (dnn_result != DNN_SUCCESS){
         av_log(ctx, AV_LOG_ERROR, "failed to execute model\n");
         av_frame_free(&in);
@@ -103,38 +100,13 @@ static int filter_frame(AVFilterLink *inlink, AVFrame *in)
 static av_cold int init(AVFilterContext *ctx)
 {
     DRContext *dr_context = ctx->priv;
-
-    dr_context->dnn_module = ff_get_dnn_module(dr_context->backend_type);
-    if (!dr_context->dnn_module) {
-        av_log(ctx, AV_LOG_ERROR, "could not create DNN module for requested backend\n");
-        return AVERROR(ENOMEM);
-    }
-    if (!dr_context->model_filename) {
-        av_log(ctx, AV_LOG_ERROR, "model file for network is not specified\n");
-        return AVERROR(EINVAL);
-    }
-    if (!dr_context->dnn_module->load_model) {
-        av_log(ctx, AV_LOG_ERROR, "load_model for network is not specified\n");
-        return AVERROR(EINVAL);
-    }
-
-    dr_context->model = (dr_context->dnn_module->load_model)(dr_context->model_filename, NULL, NULL);
-    if (!dr_context->model) {
-        av_log(ctx, AV_LOG_ERROR, "could not load DNN model\n");
-        return AVERROR(EINVAL);
-    }
-
-    return 0;
+    return ff_dnn_init(&dr_context->dnnctx, ctx);
 }
 
 static av_cold void uninit(AVFilterContext *ctx)
 {
     DRContext *dr_context = ctx->priv;
-
-    if (dr_context->dnn_module) {
-        (dr_context->dnn_module->free_model)(&dr_context->model);
-        av_freep(&dr_context->dnn_module);
-    }
+    ff_dnn_uninit(&dr_context->dnnctx);
 }
 
 static const AVFilterPad derain_inputs[] = {
diff --git a/libavfilter/vf_dnn_processing.c b/libavfilter/vf_dnn_processing.c
index 09e1699b9e..08ebf122c9 100644
--- a/libavfilter/vf_dnn_processing.c
+++ b/libavfilter/vf_dnn_processing.c
@@ -29,7 +29,7 @@
 #include "libavutil/avassert.h"
 #include "libavutil/imgutils.h"
 #include "filters.h"
-#include "dnn_interface.h"
+#include "dnn_filter_common.h"
 #include "formats.h"
 #include "internal.h"
 #include "libswscale/swscale.h"
@@ -37,22 +37,12 @@
 
 typedef struct DnnProcessingContext {
     const AVClass *class;
-
-    char *model_filename;
-    DNNBackendType backend_type;
-    char *model_inputname;
-    char *model_outputname;
-    char *backend_options;
-    int async;
-
-    DNNModule *dnn_module;
-    DNNModel *model;
-
+    DnnContext dnnctx;
     struct SwsContext *sws_uv_scale;
     int sws_uv_height;
 } DnnProcessingContext;
 
-#define OFFSET(x) offsetof(DnnProcessingContext, x)
+#define OFFSET(x) offsetof(DnnProcessingContext, dnnctx.x)
 #define FLAGS AV_OPT_FLAG_FILTERING_PARAM | AV_OPT_FLAG_VIDEO_PARAM
 static const AVOption dnn_processing_options[] = {
     { "dnn_backend", "DNN backend",                OFFSET(backend_type),     AV_OPT_TYPE_INT,       { .i64 = 0 },    INT_MIN, INT_MAX, FLAGS, "backend" },
@@ -63,11 +53,7 @@ static const AVOption dnn_processing_options[] = {
 #if (CONFIG_LIBOPENVINO == 1)
     { "openvino",    "openvino backend flag",      0,                        AV_OPT_TYPE_CONST,     { .i64 = 2 },    0, 0, FLAGS, "backend" },
 #endif
-    { "model",       "path to model file",         OFFSET(model_filename),   AV_OPT_TYPE_STRING,    { .str = NULL }, 0, 0, FLAGS },
-    { "input",       "input name of the model",    OFFSET(model_inputname),  AV_OPT_TYPE_STRING,    { .str = NULL }, 0, 0, FLAGS },
-    { "output",      "output name of the model",   OFFSET(model_outputname), AV_OPT_TYPE_STRING,    { .str = NULL }, 0, 0, FLAGS },
-    { "options",     "backend options",            OFFSET(backend_options),  AV_OPT_TYPE_STRING,    { .str = NULL }, 0, 0, FLAGS },
-    { "async",       "use DNN async inference",    OFFSET(async),            AV_OPT_TYPE_BOOL,      { .i64 = 1},     0, 1, FLAGS},
+    DNN_COMMON_OPTIONS
     { NULL }
 };
 
@@ -76,49 +62,7 @@ AVFILTER_DEFINE_CLASS(dnn_processing);
 static av_cold int init(AVFilterContext *context)
 {
     DnnProcessingContext *ctx = context->priv;
-
-    if (!ctx->model_filename) {
-        av_log(ctx, AV_LOG_ERROR, "model file for network is not specified\n");
-        return AVERROR(EINVAL);
-    }
-    if (!ctx->model_inputname) {
-        av_log(ctx, AV_LOG_ERROR, "input name of the model network is not specified\n");
-        return AVERROR(EINVAL);
-    }
-    if (!ctx->model_outputname) {
-        av_log(ctx, AV_LOG_ERROR, "output name of the model network is not specified\n");
-        return AVERROR(EINVAL);
-    }
-
-    ctx->dnn_module = ff_get_dnn_module(ctx->backend_type);
-    if (!ctx->dnn_module) {
-        av_log(ctx, AV_LOG_ERROR, "could not create DNN module for requested backend\n");
-        return AVERROR(ENOMEM);
-    }
-    if (!ctx->dnn_module->load_model) {
-        av_log(ctx, AV_LOG_ERROR, "load_model for network is not specified\n");
-        return AVERROR(EINVAL);
-    }
-
-    ctx->model = (ctx->dnn_module->load_model)(ctx->model_filename, ctx->backend_options, context);
-    if (!ctx->model) {
-        av_log(ctx, AV_LOG_ERROR, "could not load DNN model\n");
-        return AVERROR(EINVAL);
-    }
-
-    if (!ctx->dnn_module->execute_model_async && ctx->async) {
-        ctx->async = 0;
-        av_log(ctx, AV_LOG_WARNING, "this backend does not support async execution, roll back to sync.\n");
-    }
-
-#if !HAVE_PTHREAD_CANCEL
-    if (ctx->async) {
-        ctx->async = 0;
-        av_log(ctx, AV_LOG_WARNING, "pthread is not supported, roll back to sync.\n");
-    }
-#endif
-
-    return 0;
+    return ff_dnn_init(&ctx->dnnctx, context);
 }
 
 static int query_formats(AVFilterContext *context)
@@ -199,7 +143,7 @@ static int config_input(AVFilterLink *inlink)
     DNNData model_input;
     int check;
 
-    result = ctx->model->get_input(ctx->model->model, &model_input, ctx->model_inputname);
+    result = ff_dnn_get_input(&ctx->dnnctx, &model_input);
     if (result != DNN_SUCCESS) {
         av_log(ctx, AV_LOG_ERROR, "could not get input from the model\n");
         return AVERROR(EIO);
@@ -259,8 +203,7 @@ static int config_output(AVFilterLink *outlink)
     AVFilterLink *inlink = context->inputs[0];
 
     // have a try run in case that the dnn model resize the frame
-    result = ctx->model->get_output(ctx->model->model, ctx->model_inputname, inlink->w, inlink->h,
-                                    ctx->model_outputname, &outlink->w, &outlink->h);
+    result = ff_dnn_get_output(&ctx->dnnctx, inlink->w, inlink->h, &outlink->w, &outlink->h);
     if (result != DNN_SUCCESS) {
         av_log(ctx, AV_LOG_ERROR, "could not get output from the model\n");
         return AVERROR(EIO);
@@ -314,8 +257,7 @@ static int filter_frame(AVFilterLink *inlink, AVFrame *in)
     }
     av_frame_copy_props(out, in);
 
-    dnn_result = (ctx->dnn_module->execute_model)(ctx->model, ctx->model_inputname, in,
-                                                  (const char **)&ctx->model_outputname, 1, out);
+    dnn_result = ff_dnn_execute_model(&ctx->dnnctx, in, out);
     if (dnn_result != DNN_SUCCESS){
         av_log(ctx, AV_LOG_ERROR, "failed to execute model\n");
         av_frame_free(&in);
@@ -376,7 +318,7 @@ static int flush_frame(AVFilterLink *outlink, int64_t pts, int64_t *out_pts)
     int ret;
     DNNAsyncStatusType async_state;
 
-    ret = (ctx->dnn_module->flush)(ctx->model);
+    ret = ff_dnn_flush(&ctx->dnnctx);
     if (ret != DNN_SUCCESS) {
         return -1;
     }
@@ -384,7 +326,7 @@ static int flush_frame(AVFilterLink *outlink, int64_t pts, int64_t *out_pts)
     do {
         AVFrame *in_frame = NULL;
         AVFrame *out_frame = NULL;
-        async_state = (ctx->dnn_module->get_async_result)(ctx->model, &in_frame, &out_frame);
+        async_state = ff_dnn_get_async_result(&ctx->dnnctx, &in_frame, &out_frame);
         if (out_frame) {
             if (isPlanarYUV(in_frame->format))
                 copy_uv_planes(ctx, out_frame, in_frame);
@@ -405,7 +347,7 @@ static int activate_async(AVFilterContext *filter_ctx)
 {
     AVFilterLink *inlink = filter_ctx->inputs[0];
     AVFilterLink *outlink = filter_ctx->outputs[0];
-    DnnProcessingContext *ctx = (DnnProcessingContext *)filter_ctx->priv;
+    DnnProcessingContext *ctx = filter_ctx->priv;
     AVFrame *in = NULL, *out = NULL;
     int64_t pts;
     int ret, status;
@@ -426,8 +368,7 @@ static int activate_async(AVFilterContext *filter_ctx)
                 return AVERROR(ENOMEM);
             }
             av_frame_copy_props(out, in);
-            if ((ctx->dnn_module->execute_model_async)(ctx->model, ctx->model_inputname, in,
-                                                       (const char **)&ctx->model_outputname, 1, out) != DNN_SUCCESS) {
+            if (ff_dnn_execute_model_async(&ctx->dnnctx, in, out) != DNN_SUCCESS) {
                 return AVERROR(EIO);
             }
         }
@@ -437,7 +378,7 @@ static int activate_async(AVFilterContext *filter_ctx)
     do {
         AVFrame *in_frame = NULL;
         AVFrame *out_frame = NULL;
-        async_state = (ctx->dnn_module->get_async_result)(ctx->model, &in_frame, &out_frame);
+        async_state = ff_dnn_get_async_result(&ctx->dnnctx, &in_frame, &out_frame);
         if (out_frame) {
             if (isPlanarYUV(in_frame->format))
                 copy_uv_planes(ctx, out_frame, in_frame);
@@ -471,7 +412,7 @@ static int activate(AVFilterContext *filter_ctx)
 {
     DnnProcessingContext *ctx = filter_ctx->priv;
 
-    if (ctx->async)
+    if (ctx->dnnctx.async)
         return activate_async(filter_ctx);
     else
         return activate_sync(filter_ctx);
@@ -482,11 +423,7 @@ static av_cold void uninit(AVFilterContext *ctx)
     DnnProcessingContext *context = ctx->priv;
 
     sws_freeContext(context->sws_uv_scale);
-
-    if (context->dnn_module)
-        (context->dnn_module->free_model)(&context->model);
-
-    av_freep(&context->dnn_module);
+    ff_dnn_uninit(&context->dnnctx);
 }
 
 static const AVFilterPad dnn_processing_inputs[] = {
diff --git a/libavfilter/vf_sr.c b/libavfilter/vf_sr.c
index fe6c5d3c0d..20334a84c4 100644
--- a/libavfilter/vf_sr.c
+++ b/libavfilter/vf_sr.c
@@ -32,15 +32,11 @@
 #include "libavutil/pixdesc.h"
 #include "libavformat/avio.h"
 #include "libswscale/swscale.h"
-#include "dnn_interface.h"
+#include "dnn_filter_common.h"
 
 typedef struct SRContext {
     const AVClass *class;
-
-    char *model_filename;
-    DNNBackendType backend_type;
-    DNNModule *dnn_module;
-    DNNModel *model;
+    DnnContext dnnctx;
     int scale_factor;
     struct SwsContext *sws_uv_scale;
     int sws_uv_height;
@@ -50,13 +46,15 @@ typedef struct SRContext {
 #define OFFSET(x) offsetof(SRContext, x)
 #define FLAGS AV_OPT_FLAG_FILTERING_PARAM | AV_OPT_FLAG_VIDEO_PARAM
 static const AVOption sr_options[] = {
-    { "dnn_backend", "DNN backend used for model execution", OFFSET(backend_type), AV_OPT_TYPE_INT, { .i64 = 0 }, 0, 1, FLAGS, "backend" },
+    { "dnn_backend", "DNN backend used for model execution", OFFSET(dnnctx.backend_type), AV_OPT_TYPE_INT, { .i64 = 0 }, 0, 1, FLAGS, "backend" },
     { "native", "native backend flag", 0, AV_OPT_TYPE_CONST, { .i64 = 0 }, 0, 0, FLAGS, "backend" },
 #if (CONFIG_LIBTENSORFLOW == 1)
     { "tensorflow", "tensorflow backend flag", 0, AV_OPT_TYPE_CONST, { .i64 = 1 }, 0, 0, FLAGS, "backend" },
 #endif
     { "scale_factor", "scale factor for SRCNN model", OFFSET(scale_factor), AV_OPT_TYPE_INT, { .i64 = 2 }, 2, 4, FLAGS },
-    { "model", "path to model file specifying network architecture and its parameters", OFFSET(model_filename), AV_OPT_TYPE_STRING, {.str=NULL}, 0, 0, FLAGS },
+    { "model", "path to model file specifying network architecture and its parameters", OFFSET(dnnctx.model_filename), AV_OPT_TYPE_STRING, {.str=NULL}, 0, 0, FLAGS },
+    { "input",       "input name of the model",     OFFSET(dnnctx.model_inputname),  AV_OPT_TYPE_STRING,    { .str = "x" },  0, 0, FLAGS },
+    { "output",      "output name of the model",    OFFSET(dnnctx.model_outputname), AV_OPT_TYPE_STRING,    { .str = "y" },  0, 0, FLAGS },
     { NULL }
 };
 
@@ -65,28 +63,7 @@ AVFILTER_DEFINE_CLASS(sr);
 static av_cold int init(AVFilterContext *context)
 {
     SRContext *sr_context = context->priv;
-
-    sr_context->dnn_module = ff_get_dnn_module(sr_context->backend_type);
-    if (!sr_context->dnn_module){
-        av_log(context, AV_LOG_ERROR, "could not create DNN module for requested backend\n");
-        return AVERROR(ENOMEM);
-    }
-
-    if (!sr_context->model_filename){
-        av_log(context, AV_LOG_ERROR, "model file for network was not specified\n");
-        return AVERROR(EIO);
-    }
-    if (!sr_context->dnn_module->load_model) {
-        av_log(context, AV_LOG_ERROR, "load_model for network was not specified\n");
-        return AVERROR(EIO);
-    }
-    sr_context->model = (sr_context->dnn_module->load_model)(sr_context->model_filename, NULL, NULL);
-    if (!sr_context->model){
-        av_log(context, AV_LOG_ERROR, "could not load DNN model\n");
-        return AVERROR(EIO);
-    }
-
-    return 0;
+    return ff_dnn_init(&sr_context->dnnctx, context);
 }
 
 static int query_formats(AVFilterContext *context)
@@ -114,8 +91,7 @@ static int config_output(AVFilterLink *outlink)
     int out_width, out_height;
 
     // have a try run in case that the dnn model resize the frame
-    result = ctx->model->get_output(ctx->model->model, "x", inlink->w, inlink->h,
-                                    "y", &out_width, &out_height);
+    result = ff_dnn_get_output(&ctx->dnnctx, inlink->w, inlink->h, &out_width, &out_height);
     if (result != DNN_SUCCESS) {
         av_log(ctx, AV_LOG_ERROR, "could not get output from the model\n");
         return AVERROR(EIO);
@@ -155,7 +131,6 @@ static int filter_frame(AVFilterLink *inlink, AVFrame *in)
     AVFilterLink *outlink = context->outputs[0];
     AVFrame *out = ff_get_video_buffer(outlink, outlink->w, outlink->h);
     DNNReturnType dnn_result;
-    const char *model_output_name = "y";
 
     if (!out){
         av_log(context, AV_LOG_ERROR, "could not allocate memory for output frame\n");
@@ -168,11 +143,9 @@ static int filter_frame(AVFilterLink *inlink, AVFrame *in)
         sws_scale(ctx->sws_pre_scale,
                     (const uint8_t **)in->data, in->linesize, 0, in->height,
                     out->data, out->linesize);
-        dnn_result = (ctx->dnn_module->execute_model)(ctx->model, "x", out,
-                                                      (const char **)&model_output_name, 1, out);
+        dnn_result = ff_dnn_execute_model(&ctx->dnnctx, out, out);
     } else {
-        dnn_result = (ctx->dnn_module->execute_model)(ctx->model, "x", in,
-                                                      (const char **)&model_output_name, 1, out);
+        dnn_result = ff_dnn_execute_model(&ctx->dnnctx, in, out);
     }
 
     if (dnn_result != DNN_SUCCESS){
@@ -197,11 +170,7 @@ static av_cold void uninit(AVFilterContext *context)
 {
     SRContext *sr_context = context->priv;
 
-    if (sr_context->dnn_module){
-        (sr_context->dnn_module->free_model)(&sr_context->model);
-        av_freep(&sr_context->dnn_module);
-    }
-
+    ff_dnn_uninit(&sr_context->dnnctx);
     sws_freeContext(sr_context->sws_uv_scale);
     sws_freeContext(sr_context->sws_pre_scale);
 }



More information about the ffmpeg-cvslog mailing list