#include "libavutil/pixdesc.h"
#include "libavutil/avassert.h"
#include "libavutil/imgutils.h"
-#include "avfilter.h"
-#include "dnn_interface.h"
+#include "filters.h"
+#include "dnn_filter_common.h"
#include "formats.h"
#include "internal.h"
#include "libswscale/swscale.h"
+#include "libavutil/time.h"
typedef struct DnnProcessingContext {
const AVClass *class;
-
- char *model_filename;
- DNNBackendType backend_type;
- char *model_inputname;
- char *model_outputname;
- char *backend_options;
-
- 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" },
#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 },
+ DNN_COMMON_OPTIONS
{ NULL }
};
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, ctx);
- if (!ctx->model) {
- av_log(ctx, AV_LOG_ERROR, "could not load DNN model\n");
- return AVERROR(EINVAL);
- }
-
- return 0;
+ return ff_dnn_init(&ctx->dnnctx, DFT_PROCESS_FRAME, context);
}
static int query_formats(AVFilterContext *context)
AV_PIX_FMT_GRAY8, AV_PIX_FMT_GRAYF32,
AV_PIX_FMT_YUV420P, AV_PIX_FMT_YUV422P,
AV_PIX_FMT_YUV444P, AV_PIX_FMT_YUV410P, AV_PIX_FMT_YUV411P,
+ AV_PIX_FMT_NV12,
AV_PIX_FMT_NONE
};
AVFilterFormats *fmts_list = ff_make_format_list(pix_fmts);
return AVERROR(EIO);
}
if (model_input->dt != DNN_FLOAT) {
- av_log(ctx, AV_LOG_ERROR, "only support dnn models with input data type as float32.\n");
+ avpriv_report_missing_feature(ctx, "data type rather than DNN_FLOAT");
return AVERROR(EIO);
}
case AV_PIX_FMT_YUV444P:
case AV_PIX_FMT_YUV410P:
case AV_PIX_FMT_YUV411P:
+ case AV_PIX_FMT_NV12:
if (model_input->channels != 1) {
LOG_FORMAT_CHANNEL_MISMATCH();
return AVERROR(EIO);
}
return 0;
default:
- av_log(ctx, AV_LOG_ERROR, "%s not supported.\n", av_get_pix_fmt_name(fmt));
+ avpriv_report_missing_feature(ctx, "%s", av_get_pix_fmt_name(fmt));
return AVERROR(EIO);
}
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);
if (isPlanarYUV(fmt)) {
if (inlink->w != outlink->w || inlink->h != outlink->h) {
- const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(fmt);
- int sws_src_h = AV_CEIL_RSHIFT(inlink->h, desc->log2_chroma_h);
- int sws_src_w = AV_CEIL_RSHIFT(inlink->w, desc->log2_chroma_w);
- int sws_dst_h = AV_CEIL_RSHIFT(outlink->h, desc->log2_chroma_h);
- int sws_dst_w = AV_CEIL_RSHIFT(outlink->w, desc->log2_chroma_w);
- ctx->sws_uv_scale = sws_getContext(sws_src_w, sws_src_h, AV_PIX_FMT_GRAY8,
- sws_dst_w, sws_dst_h, AV_PIX_FMT_GRAY8,
- SWS_BICUBIC, NULL, NULL, NULL);
- ctx->sws_uv_height = sws_src_h;
+ if (fmt == AV_PIX_FMT_NV12) {
+ ctx->sws_uv_scale = sws_getContext(inlink->w >> 1, inlink->h >> 1, AV_PIX_FMT_YA8,
+ outlink->w >> 1, outlink->h >> 1, AV_PIX_FMT_YA8,
+ SWS_BICUBIC, NULL, NULL, NULL);
+ ctx->sws_uv_height = inlink->h >> 1;
+ } else {
+ const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(fmt);
+ int sws_src_h = AV_CEIL_RSHIFT(inlink->h, desc->log2_chroma_h);
+ int sws_src_w = AV_CEIL_RSHIFT(inlink->w, desc->log2_chroma_w);
+ int sws_dst_h = AV_CEIL_RSHIFT(outlink->h, desc->log2_chroma_h);
+ int sws_dst_w = AV_CEIL_RSHIFT(outlink->w, desc->log2_chroma_w);
+ ctx->sws_uv_scale = sws_getContext(sws_src_w, sws_src_h, AV_PIX_FMT_GRAY8,
+ sws_dst_w, sws_dst_h, AV_PIX_FMT_GRAY8,
+ SWS_BICUBIC, NULL, NULL, NULL);
+ ctx->sws_uv_height = sws_src_h;
+ }
}
}
DnnProcessingContext *ctx = context->priv;
DNNReturnType result;
AVFilterLink *inlink = context->inputs[0];
- AVFrame *out = NULL;
-
- AVFrame *fake_in = ff_get_video_buffer(inlink, inlink->w, inlink->h);
- result = (ctx->model->set_input)(ctx->model->model, fake_in, ctx->model_inputname);
- if (result != DNN_SUCCESS) {
- av_log(ctx, AV_LOG_ERROR, "could not set input for the model\n");
- return AVERROR(EIO);
- }
// have a try run in case that the dnn model resize the frame
- out = ff_get_video_buffer(inlink, inlink->w, inlink->h);
- result = (ctx->dnn_module->execute_model)(ctx->model, (const char **)&ctx->model_outputname, 1, out);
- if (result != DNN_SUCCESS){
- av_log(ctx, AV_LOG_ERROR, "failed to execute model\n");
+ 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);
}
- outlink->w = out->width;
- outlink->h = out->height;
-
- av_frame_free(&fake_in);
- av_frame_free(&out);
prepare_uv_scale(outlink);
return 0;
in->data[i], in->linesize[i],
bytewidth, uv_height);
}
+ } else if (in->format == AV_PIX_FMT_NV12) {
+ sws_scale(ctx->sws_uv_scale, (const uint8_t **)(in->data + 1), in->linesize + 1,
+ 0, ctx->sws_uv_height, out->data + 1, out->linesize + 1);
} else {
sws_scale(ctx->sws_uv_scale, (const uint8_t **)(in->data + 1), in->linesize + 1,
0, ctx->sws_uv_height, out->data + 1, out->linesize + 1);
DNNReturnType dnn_result;
AVFrame *out;
- dnn_result = (ctx->model->set_input)(ctx->model->model, in, ctx->model_inputname);
- if (dnn_result != DNN_SUCCESS) {
- av_log(ctx, AV_LOG_ERROR, "could not set input for the model\n");
- av_frame_free(&in);
- return AVERROR(EIO);
- }
-
out = ff_get_video_buffer(outlink, outlink->w, outlink->h);
if (!out) {
av_frame_free(&in);
}
av_frame_copy_props(out, in);
- dnn_result = (ctx->dnn_module->execute_model)(ctx->model, (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);
return ff_filter_frame(outlink, out);
}
+static int activate_sync(AVFilterContext *filter_ctx)
+{
+ AVFilterLink *inlink = filter_ctx->inputs[0];
+ AVFilterLink *outlink = filter_ctx->outputs[0];
+ AVFrame *in = NULL;
+ int64_t pts;
+ int ret, status;
+ int got_frame = 0;
+
+ FF_FILTER_FORWARD_STATUS_BACK(outlink, inlink);
+
+ do {
+ // drain all input frames
+ ret = ff_inlink_consume_frame(inlink, &in);
+ if (ret < 0)
+ return ret;
+ if (ret > 0) {
+ ret = filter_frame(inlink, in);
+ if (ret < 0)
+ return ret;
+ got_frame = 1;
+ }
+ } while (ret > 0);
+
+ // if frame got, schedule to next filter
+ if (got_frame)
+ return 0;
+
+ if (ff_inlink_acknowledge_status(inlink, &status, &pts)) {
+ if (status == AVERROR_EOF) {
+ ff_outlink_set_status(outlink, status, pts);
+ return ret;
+ }
+ }
+
+ FF_FILTER_FORWARD_WANTED(outlink, inlink);
+
+ return FFERROR_NOT_READY;
+}
+
+static int flush_frame(AVFilterLink *outlink, int64_t pts, int64_t *out_pts)
+{
+ DnnProcessingContext *ctx = outlink->src->priv;
+ int ret;
+ DNNAsyncStatusType async_state;
+
+ ret = ff_dnn_flush(&ctx->dnnctx);
+ if (ret != DNN_SUCCESS) {
+ return -1;
+ }
+
+ do {
+ AVFrame *in_frame = NULL;
+ AVFrame *out_frame = NULL;
+ 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);
+ av_frame_free(&in_frame);
+ ret = ff_filter_frame(outlink, out_frame);
+ if (ret < 0)
+ return ret;
+ if (out_pts)
+ *out_pts = out_frame->pts + pts;
+ }
+ av_usleep(5000);
+ } while (async_state >= DAST_NOT_READY);
+
+ return 0;
+}
+
+static int activate_async(AVFilterContext *filter_ctx)
+{
+ AVFilterLink *inlink = filter_ctx->inputs[0];
+ AVFilterLink *outlink = filter_ctx->outputs[0];
+ DnnProcessingContext *ctx = filter_ctx->priv;
+ AVFrame *in = NULL, *out = NULL;
+ int64_t pts;
+ int ret, status;
+ int got_frame = 0;
+ int async_state;
+
+ FF_FILTER_FORWARD_STATUS_BACK(outlink, inlink);
+
+ do {
+ // drain all input frames
+ ret = ff_inlink_consume_frame(inlink, &in);
+ if (ret < 0)
+ return ret;
+ if (ret > 0) {
+ out = ff_get_video_buffer(outlink, outlink->w, outlink->h);
+ if (!out) {
+ av_frame_free(&in);
+ return AVERROR(ENOMEM);
+ }
+ av_frame_copy_props(out, in);
+ if (ff_dnn_execute_model_async(&ctx->dnnctx, in, out) != DNN_SUCCESS) {
+ return AVERROR(EIO);
+ }
+ }
+ } while (ret > 0);
+
+ // drain all processed frames
+ do {
+ AVFrame *in_frame = NULL;
+ AVFrame *out_frame = NULL;
+ 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);
+ av_frame_free(&in_frame);
+ ret = ff_filter_frame(outlink, out_frame);
+ if (ret < 0)
+ return ret;
+ got_frame = 1;
+ }
+ } while (async_state == DAST_SUCCESS);
+
+ // if frame got, schedule to next filter
+ if (got_frame)
+ return 0;
+
+ if (ff_inlink_acknowledge_status(inlink, &status, &pts)) {
+ if (status == AVERROR_EOF) {
+ int64_t out_pts = pts;
+ ret = flush_frame(outlink, pts, &out_pts);
+ ff_outlink_set_status(outlink, status, out_pts);
+ return ret;
+ }
+ }
+
+ FF_FILTER_FORWARD_WANTED(outlink, inlink);
+
+ return 0;
+}
+
+static int activate(AVFilterContext *filter_ctx)
+{
+ DnnProcessingContext *ctx = filter_ctx->priv;
+
+ if (ctx->dnnctx.async)
+ return activate_async(filter_ctx);
+ else
+ return activate_sync(filter_ctx);
+}
+
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[] = {
.name = "default",
.type = AVMEDIA_TYPE_VIDEO,
.config_props = config_input,
- .filter_frame = filter_frame,
},
{ NULL }
};
{ NULL }
};
-AVFilter ff_vf_dnn_processing = {
+const AVFilter ff_vf_dnn_processing = {
.name = "dnn_processing",
.description = NULL_IF_CONFIG_SMALL("Apply DNN processing filter to the input."),
.priv_size = sizeof(DnnProcessingContext),
.inputs = dnn_processing_inputs,
.outputs = dnn_processing_outputs,
.priv_class = &dnn_processing_class,
+ .activate = activate,
};
projects / ffmpeg / blobdiff