2018-06-13 21:37:12 +00:00
|
|
|
/*
|
|
|
|
* Copyright (c) 2018 Sergey Lavrushkin
|
|
|
|
*
|
|
|
|
* 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
|
|
|
|
* Filter implementing image super-resolution using deep convolutional networks.
|
|
|
|
* https://arxiv.org/abs/1501.00092
|
|
|
|
* https://arxiv.org/abs/1609.05158
|
|
|
|
*/
|
|
|
|
|
|
|
|
#include "avfilter.h"
|
|
|
|
#include "formats.h"
|
|
|
|
#include "internal.h"
|
|
|
|
#include "libavutil/opt.h"
|
2019-06-01 03:20:33 +00:00
|
|
|
#include "libavutil/pixdesc.h"
|
2018-06-13 21:37:12 +00:00
|
|
|
#include "libavformat/avio.h"
|
|
|
|
#include "libswscale/swscale.h"
|
|
|
|
#include "dnn_interface.h"
|
|
|
|
|
|
|
|
typedef struct SRContext {
|
|
|
|
const AVClass *class;
|
|
|
|
|
2018-07-27 16:34:02 +00:00
|
|
|
char *model_filename;
|
2018-06-13 21:37:12 +00:00
|
|
|
DNNBackendType backend_type;
|
2018-07-27 16:34:02 +00:00
|
|
|
DNNModule *dnn_module;
|
|
|
|
DNNModel *model;
|
2019-04-25 02:14:42 +00:00
|
|
|
DNNInputData input;
|
|
|
|
DNNData output;
|
2018-06-13 21:37:12 +00:00
|
|
|
int scale_factor;
|
2018-07-31 15:58:28 +00:00
|
|
|
struct SwsContext *sws_contexts[3];
|
|
|
|
int sws_slice_h, sws_input_linesize, sws_output_linesize;
|
2018-06-13 21:37:12 +00:00
|
|
|
} SRContext;
|
|
|
|
|
|
|
|
#define OFFSET(x) offsetof(SRContext, x)
|
|
|
|
#define FLAGS AV_OPT_FLAG_FILTERING_PARAM | AV_OPT_FLAG_VIDEO_PARAM
|
|
|
|
static const AVOption sr_options[] = {
|
2019-05-25 11:59:42 +00:00
|
|
|
{ "dnn_backend", "DNN backend used for model execution", OFFSET(backend_type), AV_OPT_TYPE_INT, { .i64 = 0 }, 0, 1, FLAGS, "backend" },
|
2018-06-13 21:37:12 +00:00
|
|
|
{ "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
|
2018-09-06 11:33:06 +00:00
|
|
|
{ "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 },
|
2018-06-13 21:37:12 +00:00
|
|
|
{ NULL }
|
|
|
|
};
|
|
|
|
|
|
|
|
AVFILTER_DEFINE_CLASS(sr);
|
|
|
|
|
2018-07-27 16:34:02 +00:00
|
|
|
static av_cold int init(AVFilterContext *context)
|
2018-06-13 21:37:12 +00:00
|
|
|
{
|
2018-07-27 16:34:02 +00:00
|
|
|
SRContext *sr_context = context->priv;
|
2018-06-13 21:37:12 +00:00
|
|
|
|
|
|
|
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);
|
|
|
|
}
|
2019-06-01 03:28:56 +00:00
|
|
|
|
2018-06-13 21:37:12 +00:00
|
|
|
if (!sr_context->model_filename){
|
2018-09-06 11:33:06 +00:00
|
|
|
av_log(context, AV_LOG_ERROR, "model file for network was not specified\n");
|
|
|
|
return AVERROR(EIO);
|
2018-06-13 21:37:12 +00:00
|
|
|
}
|
2019-06-01 03:28:56 +00:00
|
|
|
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);
|
2018-06-13 21:37:12 +00:00
|
|
|
if (!sr_context->model){
|
|
|
|
av_log(context, AV_LOG_ERROR, "could not load DNN model\n");
|
|
|
|
return AVERROR(EIO);
|
|
|
|
}
|
|
|
|
|
2019-04-25 02:14:42 +00:00
|
|
|
sr_context->input.dt = DNN_FLOAT;
|
2018-07-31 15:58:28 +00:00
|
|
|
sr_context->sws_contexts[0] = NULL;
|
|
|
|
sr_context->sws_contexts[1] = NULL;
|
|
|
|
sr_context->sws_contexts[2] = NULL;
|
|
|
|
|
2018-06-13 21:37:12 +00:00
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2018-07-27 16:34:02 +00:00
|
|
|
static int query_formats(AVFilterContext *context)
|
2018-06-13 21:37:12 +00:00
|
|
|
{
|
|
|
|
const enum AVPixelFormat pixel_formats[] = {AV_PIX_FMT_YUV420P, AV_PIX_FMT_YUV422P, AV_PIX_FMT_YUV444P,
|
|
|
|
AV_PIX_FMT_YUV410P, AV_PIX_FMT_YUV411P, AV_PIX_FMT_GRAY8,
|
|
|
|
AV_PIX_FMT_NONE};
|
2018-07-27 16:34:02 +00:00
|
|
|
AVFilterFormats *formats_list;
|
2018-06-13 21:37:12 +00:00
|
|
|
|
|
|
|
formats_list = ff_make_format_list(pixel_formats);
|
|
|
|
if (!formats_list){
|
|
|
|
av_log(context, AV_LOG_ERROR, "could not create formats list\n");
|
|
|
|
return AVERROR(ENOMEM);
|
|
|
|
}
|
2018-07-31 15:58:28 +00:00
|
|
|
|
2018-06-13 21:37:12 +00:00
|
|
|
return ff_set_common_formats(context, formats_list);
|
|
|
|
}
|
|
|
|
|
2018-07-27 16:34:02 +00:00
|
|
|
static int config_props(AVFilterLink *inlink)
|
2018-06-13 21:37:12 +00:00
|
|
|
{
|
2018-07-27 16:34:02 +00:00
|
|
|
AVFilterContext *context = inlink->dst;
|
|
|
|
SRContext *sr_context = context->priv;
|
|
|
|
AVFilterLink *outlink = context->outputs[0];
|
2018-06-13 21:37:12 +00:00
|
|
|
DNNReturnType result;
|
|
|
|
int sws_src_h, sws_src_w, sws_dst_h, sws_dst_w;
|
2019-04-25 02:14:33 +00:00
|
|
|
const char *model_output_name = "y";
|
2018-06-13 21:37:12 +00:00
|
|
|
|
2018-09-06 11:33:06 +00:00
|
|
|
sr_context->input.width = inlink->w * sr_context->scale_factor;
|
|
|
|
sr_context->input.height = inlink->h * sr_context->scale_factor;
|
2018-06-13 21:37:12 +00:00
|
|
|
sr_context->input.channels = 1;
|
|
|
|
|
2019-04-25 02:14:33 +00:00
|
|
|
result = (sr_context->model->set_input_output)(sr_context->model->model, &sr_context->input, "x", &model_output_name, 1);
|
2018-06-13 21:37:12 +00:00
|
|
|
if (result != DNN_SUCCESS){
|
|
|
|
av_log(context, AV_LOG_ERROR, "could not set input and output for the model\n");
|
|
|
|
return AVERROR(EIO);
|
|
|
|
}
|
2019-04-25 02:14:01 +00:00
|
|
|
|
2019-04-25 02:14:33 +00:00
|
|
|
result = (sr_context->dnn_module->execute_model)(sr_context->model, &sr_context->output, 1);
|
2019-04-25 02:14:17 +00:00
|
|
|
if (result != DNN_SUCCESS){
|
|
|
|
av_log(context, AV_LOG_ERROR, "failed to execute loaded model\n");
|
|
|
|
return AVERROR(EIO);
|
|
|
|
}
|
|
|
|
|
2019-04-25 02:14:01 +00:00
|
|
|
if (sr_context->input.height != sr_context->output.height || sr_context->input.width != sr_context->output.width){
|
|
|
|
sr_context->input.width = inlink->w;
|
|
|
|
sr_context->input.height = inlink->h;
|
2019-04-25 02:14:33 +00:00
|
|
|
result = (sr_context->model->set_input_output)(sr_context->model->model, &sr_context->input, "x", &model_output_name, 1);
|
2019-04-25 02:14:01 +00:00
|
|
|
if (result != DNN_SUCCESS){
|
|
|
|
av_log(context, AV_LOG_ERROR, "could not set input and output for the model\n");
|
|
|
|
return AVERROR(EIO);
|
2018-09-06 11:33:06 +00:00
|
|
|
}
|
2019-04-25 02:14:33 +00:00
|
|
|
result = (sr_context->dnn_module->execute_model)(sr_context->model, &sr_context->output, 1);
|
2019-04-25 02:14:17 +00:00
|
|
|
if (result != DNN_SUCCESS){
|
|
|
|
av_log(context, AV_LOG_ERROR, "failed to execute loaded model\n");
|
|
|
|
return AVERROR(EIO);
|
|
|
|
}
|
2019-04-25 02:14:01 +00:00
|
|
|
sr_context->scale_factor = 0;
|
|
|
|
}
|
|
|
|
outlink->h = sr_context->output.height;
|
|
|
|
outlink->w = sr_context->output.width;
|
|
|
|
sr_context->sws_contexts[1] = sws_getContext(sr_context->input.width, sr_context->input.height, AV_PIX_FMT_GRAY8,
|
|
|
|
sr_context->input.width, sr_context->input.height, AV_PIX_FMT_GRAYF32,
|
|
|
|
0, NULL, NULL, NULL);
|
|
|
|
sr_context->sws_input_linesize = sr_context->input.width << 2;
|
|
|
|
sr_context->sws_contexts[2] = sws_getContext(sr_context->output.width, sr_context->output.height, AV_PIX_FMT_GRAYF32,
|
|
|
|
sr_context->output.width, sr_context->output.height, AV_PIX_FMT_GRAY8,
|
|
|
|
0, NULL, NULL, NULL);
|
|
|
|
sr_context->sws_output_linesize = sr_context->output.width << 2;
|
|
|
|
if (!sr_context->sws_contexts[1] || !sr_context->sws_contexts[2]){
|
|
|
|
av_log(context, AV_LOG_ERROR, "could not create SwsContext for conversions\n");
|
|
|
|
return AVERROR(ENOMEM);
|
|
|
|
}
|
|
|
|
if (sr_context->scale_factor){
|
|
|
|
sr_context->sws_contexts[0] = sws_getContext(inlink->w, inlink->h, inlink->format,
|
|
|
|
outlink->w, outlink->h, outlink->format,
|
|
|
|
SWS_BICUBIC, NULL, NULL, NULL);
|
|
|
|
if (!sr_context->sws_contexts[0]){
|
|
|
|
av_log(context, AV_LOG_ERROR, "could not create SwsContext for scaling\n");
|
2018-07-31 15:58:28 +00:00
|
|
|
return AVERROR(ENOMEM);
|
|
|
|
}
|
2019-04-25 02:14:01 +00:00
|
|
|
sr_context->sws_slice_h = inlink->h;
|
|
|
|
} else {
|
|
|
|
if (inlink->format != AV_PIX_FMT_GRAY8){
|
|
|
|
sws_src_h = sr_context->input.height;
|
|
|
|
sws_src_w = sr_context->input.width;
|
|
|
|
sws_dst_h = sr_context->output.height;
|
|
|
|
sws_dst_w = sr_context->output.width;
|
|
|
|
|
|
|
|
switch (inlink->format){
|
|
|
|
case AV_PIX_FMT_YUV420P:
|
|
|
|
sws_src_h = AV_CEIL_RSHIFT(sws_src_h, 1);
|
|
|
|
sws_src_w = AV_CEIL_RSHIFT(sws_src_w, 1);
|
|
|
|
sws_dst_h = AV_CEIL_RSHIFT(sws_dst_h, 1);
|
|
|
|
sws_dst_w = AV_CEIL_RSHIFT(sws_dst_w, 1);
|
|
|
|
break;
|
|
|
|
case AV_PIX_FMT_YUV422P:
|
|
|
|
sws_src_w = AV_CEIL_RSHIFT(sws_src_w, 1);
|
|
|
|
sws_dst_w = AV_CEIL_RSHIFT(sws_dst_w, 1);
|
|
|
|
break;
|
|
|
|
case AV_PIX_FMT_YUV444P:
|
|
|
|
break;
|
|
|
|
case AV_PIX_FMT_YUV410P:
|
|
|
|
sws_src_h = AV_CEIL_RSHIFT(sws_src_h, 2);
|
|
|
|
sws_src_w = AV_CEIL_RSHIFT(sws_src_w, 2);
|
|
|
|
sws_dst_h = AV_CEIL_RSHIFT(sws_dst_h, 2);
|
|
|
|
sws_dst_w = AV_CEIL_RSHIFT(sws_dst_w, 2);
|
|
|
|
break;
|
|
|
|
case AV_PIX_FMT_YUV411P:
|
|
|
|
sws_src_w = AV_CEIL_RSHIFT(sws_src_w, 2);
|
|
|
|
sws_dst_w = AV_CEIL_RSHIFT(sws_dst_w, 2);
|
|
|
|
break;
|
|
|
|
default:
|
2019-06-01 03:20:33 +00:00
|
|
|
av_log(context, AV_LOG_ERROR,
|
|
|
|
"could not create SwsContext for scaling for given input pixel format: %s\n",
|
|
|
|
av_get_pix_fmt_name(inlink->format));
|
2019-04-25 02:14:01 +00:00
|
|
|
return AVERROR(EIO);
|
|
|
|
}
|
|
|
|
sr_context->sws_contexts[0] = sws_getContext(sws_src_w, sws_src_h, AV_PIX_FMT_GRAY8,
|
|
|
|
sws_dst_w, sws_dst_h, AV_PIX_FMT_GRAY8,
|
2018-07-31 15:58:28 +00:00
|
|
|
SWS_BICUBIC, NULL, NULL, NULL);
|
|
|
|
if (!sr_context->sws_contexts[0]){
|
|
|
|
av_log(context, AV_LOG_ERROR, "could not create SwsContext for scaling\n");
|
2018-06-13 21:37:12 +00:00
|
|
|
return AVERROR(ENOMEM);
|
|
|
|
}
|
2019-04-25 02:14:01 +00:00
|
|
|
sr_context->sws_slice_h = sws_src_h;
|
2018-06-13 21:37:12 +00:00
|
|
|
}
|
|
|
|
}
|
2019-04-25 02:14:01 +00:00
|
|
|
|
|
|
|
return 0;
|
2018-06-13 21:37:12 +00:00
|
|
|
}
|
|
|
|
|
2018-07-27 16:34:02 +00:00
|
|
|
static int filter_frame(AVFilterLink *inlink, AVFrame *in)
|
2018-06-13 21:37:12 +00:00
|
|
|
{
|
2018-07-27 16:34:02 +00:00
|
|
|
AVFilterContext *context = inlink->dst;
|
|
|
|
SRContext *sr_context = context->priv;
|
|
|
|
AVFilterLink *outlink = context->outputs[0];
|
|
|
|
AVFrame *out = ff_get_video_buffer(outlink, outlink->w, outlink->h);
|
2018-06-13 21:37:12 +00:00
|
|
|
DNNReturnType dnn_result;
|
|
|
|
|
|
|
|
if (!out){
|
|
|
|
av_log(context, AV_LOG_ERROR, "could not allocate memory for output frame\n");
|
|
|
|
av_frame_free(&in);
|
|
|
|
return AVERROR(ENOMEM);
|
|
|
|
}
|
|
|
|
av_frame_copy_props(out, in);
|
|
|
|
out->height = sr_context->output.height;
|
|
|
|
out->width = sr_context->output.width;
|
2018-09-06 11:33:06 +00:00
|
|
|
if (sr_context->scale_factor){
|
2018-07-31 15:58:28 +00:00
|
|
|
sws_scale(sr_context->sws_contexts[0], (const uint8_t **)in->data, in->linesize,
|
2018-06-13 21:37:12 +00:00
|
|
|
0, sr_context->sws_slice_h, out->data, out->linesize);
|
2018-07-31 15:58:28 +00:00
|
|
|
|
|
|
|
sws_scale(sr_context->sws_contexts[1], (const uint8_t **)out->data, out->linesize,
|
2018-09-19 02:55:11 +00:00
|
|
|
0, out->height, (uint8_t * const*)(&sr_context->input.data),
|
|
|
|
(const int [4]){sr_context->sws_input_linesize, 0, 0, 0});
|
2019-04-25 02:14:01 +00:00
|
|
|
} else {
|
2018-07-31 15:58:28 +00:00
|
|
|
if (sr_context->sws_contexts[0]){
|
|
|
|
sws_scale(sr_context->sws_contexts[0], (const uint8_t **)(in->data + 1), in->linesize + 1,
|
2018-06-13 21:37:12 +00:00
|
|
|
0, sr_context->sws_slice_h, out->data + 1, out->linesize + 1);
|
2018-07-31 15:58:28 +00:00
|
|
|
sws_scale(sr_context->sws_contexts[0], (const uint8_t **)(in->data + 2), in->linesize + 2,
|
2018-06-13 21:37:12 +00:00
|
|
|
0, sr_context->sws_slice_h, out->data + 2, out->linesize + 2);
|
|
|
|
}
|
|
|
|
|
2018-07-31 15:58:28 +00:00
|
|
|
sws_scale(sr_context->sws_contexts[1], (const uint8_t **)in->data, in->linesize,
|
2018-09-19 02:55:11 +00:00
|
|
|
0, in->height, (uint8_t * const*)(&sr_context->input.data),
|
|
|
|
(const int [4]){sr_context->sws_input_linesize, 0, 0, 0});
|
2018-07-31 15:58:28 +00:00
|
|
|
}
|
2018-06-13 21:37:12 +00:00
|
|
|
av_frame_free(&in);
|
|
|
|
|
2019-04-25 02:14:33 +00:00
|
|
|
dnn_result = (sr_context->dnn_module->execute_model)(sr_context->model, &sr_context->output, 1);
|
2018-06-13 21:37:12 +00:00
|
|
|
if (dnn_result != DNN_SUCCESS){
|
|
|
|
av_log(context, AV_LOG_ERROR, "failed to execute loaded model\n");
|
|
|
|
return AVERROR(EIO);
|
|
|
|
}
|
|
|
|
|
2018-09-20 14:48:20 +00:00
|
|
|
sws_scale(sr_context->sws_contexts[2], (const uint8_t *[4]){(const uint8_t *)sr_context->output.data, 0, 0, 0},
|
2018-09-19 02:55:11 +00:00
|
|
|
(const int[4]){sr_context->sws_output_linesize, 0, 0, 0},
|
2018-07-31 15:58:28 +00:00
|
|
|
0, out->height, (uint8_t * const*)out->data, out->linesize);
|
2018-06-13 21:37:12 +00:00
|
|
|
|
|
|
|
return ff_filter_frame(outlink, out);
|
|
|
|
}
|
|
|
|
|
2018-07-27 16:34:02 +00:00
|
|
|
static av_cold void uninit(AVFilterContext *context)
|
2018-06-13 21:37:12 +00:00
|
|
|
{
|
2018-07-31 15:58:28 +00:00
|
|
|
int i;
|
2018-07-27 16:34:02 +00:00
|
|
|
SRContext *sr_context = context->priv;
|
2018-06-13 21:37:12 +00:00
|
|
|
|
|
|
|
if (sr_context->dnn_module){
|
|
|
|
(sr_context->dnn_module->free_model)(&sr_context->model);
|
|
|
|
av_freep(&sr_context->dnn_module);
|
|
|
|
}
|
|
|
|
|
2018-07-31 15:58:28 +00:00
|
|
|
for (i = 0; i < 3; ++i){
|
2019-06-01 03:25:46 +00:00
|
|
|
sws_freeContext(sr_context->sws_contexts[i]);
|
2018-06-13 21:37:12 +00:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
static const AVFilterPad sr_inputs[] = {
|
|
|
|
{
|
|
|
|
.name = "default",
|
|
|
|
.type = AVMEDIA_TYPE_VIDEO,
|
|
|
|
.config_props = config_props,
|
|
|
|
.filter_frame = filter_frame,
|
|
|
|
},
|
|
|
|
{ NULL }
|
|
|
|
};
|
|
|
|
|
|
|
|
static const AVFilterPad sr_outputs[] = {
|
|
|
|
{
|
|
|
|
.name = "default",
|
|
|
|
.type = AVMEDIA_TYPE_VIDEO,
|
|
|
|
},
|
|
|
|
{ NULL }
|
|
|
|
};
|
|
|
|
|
|
|
|
AVFilter ff_vf_sr = {
|
|
|
|
.name = "sr",
|
|
|
|
.description = NULL_IF_CONFIG_SMALL("Apply DNN-based image super resolution to the input."),
|
|
|
|
.priv_size = sizeof(SRContext),
|
|
|
|
.init = init,
|
|
|
|
.uninit = uninit,
|
|
|
|
.query_formats = query_formats,
|
|
|
|
.inputs = sr_inputs,
|
|
|
|
.outputs = sr_outputs,
|
|
|
|
.priv_class = &sr_class,
|
2019-05-25 12:09:41 +00:00
|
|
|
.flags = AVFILTER_FLAG_SUPPORT_TIMELINE_GENERIC,
|
2018-06-13 21:37:12 +00:00
|
|
|
};
|