45 #define OFFSET(x) offsetof(SRCNNContext, x)
46 #define FLAGS AV_OPT_FLAG_FILTERING_PARAM | AV_OPT_FLAG_VIDEO_PARAM
48 {
"dnn_backend",
"DNN backend used for model execution",
OFFSET(backend_type),
AV_OPT_TYPE_FLAGS, { .i64 = 0 }, 0, 1,
FLAGS,
"backend" },
50 #if (CONFIG_LIBTENSORFLOW == 1)
53 {
"model_filename",
"path to model file specifying network architecture and its parameters",
OFFSET(model_filename),
AV_OPT_TYPE_STRING, {.str=
NULL}, 0, 0,
FLAGS },
69 av_log(context,
AV_LOG_VERBOSE,
"model file for network was not specified, using default network for x2 upsampling\n");
75 if (!srcnn_context->
model){
134 const int slice_start = (td->
height * jobnr ) / nb_jobs;
140 for (y = slice_start; y <
slice_end; ++y){
141 for (x = 0; x < td->
width; ++x){
142 dst[x] = (float)src[x] / 255.0
f;
155 const int slice_start = (td->
height * jobnr ) / nb_jobs;
161 for (y = slice_start; y <
slice_end; ++y){
162 for (x = 0; x < td->
width; ++x){
240 .description =
NULL_IF_CONFIG_SMALL(
"Apply super resolution convolutional neural network to the input. Use bicubic upsamping with corresponding scaling factor before."),
247 .priv_class = &srcnn_class,
static int float_to_uint8(AVFilterContext *context, void *arg, int jobnr, int nb_jobs)
This structure describes decoded (raw) audio or video data.
planar YUV 4:4:4, 24bpp, (1 Cr & Cb sample per 1x1 Y samples)
static const AVFilterPad srcnn_inputs[]
Main libavfilter public API header.
int h
agreed upon image height
AVFrame * ff_get_video_buffer(AVFilterLink *link, int w, int h)
Request a picture buffer with a specific set of permissions.
#define AVFILTER_FLAG_SUPPORT_TIMELINE_GENERIC
Some filters support a generic "enable" expression option that can be used to enable or disable a fil...
const char * name
Pad name.
int ff_filter_frame(AVFilterLink *link, AVFrame *frame)
Send a frame of data to the next filter.
ptrdiff_t out_linesize[3]
#define AV_LOG_VERBOSE
Detailed information.
#define flags(name, subs,...)
A filter pad used for either input or output.
static int uint8_to_float(AVFilterContext *context, void *arg, int jobnr, int nb_jobs)
A link between two filters.
#define AV_LOG_ERROR
Something went wrong and cannot losslessly be recovered.
AVFILTER_DEFINE_CLASS(srcnn)
void av_frame_free(AVFrame **frame)
Free the frame and any dynamically allocated objects in it, e.g.
#define NULL_IF_CONFIG_SMALL(x)
Return NULL if CONFIG_SMALL is true, otherwise the argument without modification. ...
void * priv
private data for use by the filter
DNNReturnType(* execute_model)(const DNNModel *model)
#define AVFILTER_FLAG_SLICE_THREADS
The filter supports multithreading by splitting frames into multiple parts and processing them concur...
int av_frame_copy(AVFrame *dst, const AVFrame *src)
Copy the frame data from src to dst.
planar YUV 4:2:2, 16bpp, (1 Cr & Cb sample per 2x1 Y samples)
int w
agreed upon image width
static av_cold void uninit(AVFilterContext *context)
int ff_filter_get_nb_threads(AVFilterContext *ctx)
Get number of threads for current filter instance.
DNNModel *(* load_model)(const char *model_filename)
static av_cold int init(AVFilterContext *context)
DNN inference engine interface.
static const AVFilterPad inputs[]
static const AVFilterPad outputs[]
int linesize[AV_NUM_DATA_POINTERS]
For video, size in bytes of each picture line.
static const AVFilterPad srcnn_outputs[]
uint8_t pi<< 24) CONV_FUNC_GROUP(AV_SAMPLE_FMT_FLT, float, AV_SAMPLE_FMT_U8, uint8_t,(*(constuint8_t *) pi-0x80)*(1.0f/(1<< 7))) CONV_FUNC_GROUP(AV_SAMPLE_FMT_DBL, double, AV_SAMPLE_FMT_U8, uint8_t,(*(constuint8_t *) pi-0x80)*(1.0/(1<< 7))) CONV_FUNC_GROUP(AV_SAMPLE_FMT_U8, uint8_t, AV_SAMPLE_FMT_S16, int16_t,(*(constint16_t *) pi >>8)+0x80) CONV_FUNC_GROUP(AV_SAMPLE_FMT_FLT, float, AV_SAMPLE_FMT_S16, int16_t,*(constint16_t *) pi *(1.0f/(1<< 15))) CONV_FUNC_GROUP(AV_SAMPLE_FMT_DBL, double, AV_SAMPLE_FMT_S16, int16_t,*(constint16_t *) pi *(1.0/(1<< 15))) CONV_FUNC_GROUP(AV_SAMPLE_FMT_U8, uint8_t, AV_SAMPLE_FMT_S32, int32_t,(*(constint32_t *) pi >>24)+0x80) CONV_FUNC_GROUP(AV_SAMPLE_FMT_FLT, float, AV_SAMPLE_FMT_S32, int32_t,*(constint32_t *) pi *(1.0f/(1U<< 31))) CONV_FUNC_GROUP(AV_SAMPLE_FMT_DBL, double, AV_SAMPLE_FMT_S32, int32_t,*(constint32_t *) pi *(1.0/(1U<< 31))) CONV_FUNC_GROUP(AV_SAMPLE_FMT_U8, uint8_t, AV_SAMPLE_FMT_FLT, float, av_clip_uint8(lrintf(*(constfloat *) pi *(1<< 7))+0x80)) CONV_FUNC_GROUP(AV_SAMPLE_FMT_S16, int16_t, AV_SAMPLE_FMT_FLT, float, av_clip_int16(lrintf(*(constfloat *) pi *(1<< 15)))) CONV_FUNC_GROUP(AV_SAMPLE_FMT_S32, int32_t, AV_SAMPLE_FMT_FLT, float, av_clipl_int32(llrintf(*(constfloat *) pi *(1U<< 31)))) CONV_FUNC_GROUP(AV_SAMPLE_FMT_U8, uint8_t, AV_SAMPLE_FMT_DBL, double, av_clip_uint8(lrint(*(constdouble *) pi *(1<< 7))+0x80)) CONV_FUNC_GROUP(AV_SAMPLE_FMT_S16, int16_t, AV_SAMPLE_FMT_DBL, double, av_clip_int16(lrint(*(constdouble *) pi *(1<< 15)))) CONV_FUNC_GROUP(AV_SAMPLE_FMT_S32, int32_t, AV_SAMPLE_FMT_DBL, double, av_clipl_int32(llrint(*(constdouble *) pi *(1U<< 31))))#defineSET_CONV_FUNC_GROUP(ofmt, ifmt) staticvoidset_generic_function(AudioConvert *ac){}voidff_audio_convert_free(AudioConvert **ac){if(!*ac) return;ff_dither_free(&(*ac) ->dc);av_freep(ac);}AudioConvert *ff_audio_convert_alloc(AVAudioResampleContext *avr, enumAVSampleFormatout_fmt, enumAVSampleFormatin_fmt, intchannels, intsample_rate, intapply_map){AudioConvert *ac;intin_planar, out_planar;ac=av_mallocz(sizeof(*ac));if(!ac) returnNULL;ac->avr=avr;ac->out_fmt=out_fmt;ac->in_fmt=in_fmt;ac->channels=channels;ac->apply_map=apply_map;if(avr->dither_method!=AV_RESAMPLE_DITHER_NONE &&av_get_packed_sample_fmt(out_fmt)==AV_SAMPLE_FMT_S16 &&av_get_bytes_per_sample(in_fmt)>2){ac->dc=ff_dither_alloc(avr, out_fmt, in_fmt, channels, sample_rate, apply_map);if(!ac->dc){av_free(ac);returnNULL;}returnac;}in_planar=ff_sample_fmt_is_planar(in_fmt, channels);out_planar=ff_sample_fmt_is_planar(out_fmt, channels);if(in_planar==out_planar){ac->func_type=CONV_FUNC_TYPE_FLAT;ac->planes=in_planar?ac->channels:1;}elseif(in_planar) ac->func_type=CONV_FUNC_TYPE_INTERLEAVE;elseac->func_type=CONV_FUNC_TYPE_DEINTERLEAVE;set_generic_function(ac);if(ARCH_AARCH64) ff_audio_convert_init_aarch64(ac);if(ARCH_ARM) ff_audio_convert_init_arm(ac);if(ARCH_X86) ff_audio_convert_init_x86(ac);returnac;}intff_audio_convert(AudioConvert *ac, AudioData *out, AudioData *in){intuse_generic=1;intlen=in->nb_samples;intp;if(ac->dc){av_log(ac->avr, AV_LOG_TRACE,"%dsamples-audio_convert:%sto%s(dithered)\n", len, av_get_sample_fmt_name(ac->in_fmt), av_get_sample_fmt_name(ac->out_fmt));returnff_convert_dither(ac-> in
planar YUV 4:1:0, 9bpp, (1 Cr & Cb sample per 4x4 Y samples)
Describe the class of an AVClass context structure.
const char * name
Filter name.
static const AVOption srcnn_options[]
AVFilterLink ** outputs
array of pointers to output links
DNNBackendType backend_type
DNNModule * ff_get_dnn_module(DNNBackendType backend_type)
void(* free_model)(DNNModel **model)
AVFilterInternal * internal
An opaque struct for libavfilter internal use.
uint8_t * data[AV_NUM_DATA_POINTERS]
pointer to the picture/channel planes.
planar YUV 4:2:0, 12bpp, (1 Cr & Cb sample per 2x2 Y samples)
static int config_props(AVFilterLink *inlink)
static int query_formats(AVFilterContext *context)
planar YUV 4:1:1, 12bpp, (1 Cr & Cb sample per 4x1 Y samples)
avfilter_execute_func * execute
static int slice_end(AVCodecContext *avctx, AVFrame *pict)
Handle slice ends.
AVFilterContext * dst
dest filter
static int filter_frame(AVFilterLink *inlink, AVFrame *in)
DNNModel *(* load_default_model)(DNNDefaultModel model_type)
DNNReturnType(* set_input_output)(void *model, const DNNData *input, const DNNData *output)
AVPixelFormat
Pixel format.
int av_frame_copy_props(AVFrame *dst, const AVFrame *src)
Copy only "metadata" fields from src to dst.