vf_dnn_detect.c File Reference

implementing an object detecting filter using deep learning networks. More...

#include "libavutil/common.h"
#include "libavutil/file_open.h"
#include "libavutil/mem.h"
#include "libavutil/opt.h"
#include "filters.h"
#include "dnn_filter_common.h"
#include "video.h"
#include "libavutil/time.h"
#include "libavutil/avstring.h"
#include "libavutil/detection_bbox.h"
#include "libavutil/fifo.h"
#include <float.h>

Go to the source code of this file.

Data Structures
struct	DnnDetectContext

Macros
#define	OFFSET(x)
#define	OFFSET2(x)
#define	FLAGS   AV_OPT_FLAG_FILTERING_PARAM | AV_OPT_FLAG_VIDEO_PARAM

Enumerations
enum	DNNDetectionModelType { DDMT_SSD , DDMT_YOLOV1V2 , DDMT_YOLOV3 , DDMT_YOLOV4 }

Functions
	AVFILTER_DNN_DEFINE_CLASS (dnn_detect, DNN_TF|DNN_OV|DNN_ONNX)
static float	sigmoid (float x)
static float	linear (float x)
static int	dnn_detect_get_label_id (int nb_classes, int cell_size, float *label_data)
static int	dnn_detect_label_id_from_float (float f)
static int	dnn_detect_float_to_coord (double v)
static int	dnn_detect_confidence_num (double conf)
static void	dnn_detect_set_bbox_edges (AVDetectionBBox *bbox, double x0, double y0, double x1, double y1, int frame_w, int frame_h)
static float	dnn_detect_IOU (AVDetectionBBox *bbox1, AVDetectionBBox *bbox2)
static int	dnn_detect_parse_yolo_output (AVFrame *frame, DNNData *output, int output_index, AVFilterContext *filter_ctx, int64_t *anchor_used)
static int	dnn_detect_fill_side_data (AVFrame *frame, AVFilterContext *filter_ctx)
static int	dnn_detect_post_proc_yolo (AVFrame *frame, DNNData *output, AVFilterContext *filter_ctx)
static int	dnn_detect_post_proc_yolov3 (AVFrame *frame, DNNData *output, AVFilterContext *filter_ctx, int nb_outputs)
static int	dnn_detect_post_proc_ssd (AVFrame *frame, DNNData *output, int nb_outputs, AVFilterContext *filter_ctx)
static int	dnn_detect_post_proc_anchored (AVFrame *frame, DNNData *output, int nb_outputs, AVFilterContext *filter_ctx)
static int	dnn_detect_post_proc_tf (AVFrame *frame, DNNData *output, AVFilterContext *filter_ctx)
static int	dnn_detect_post_proc (AVFrame *frame, DNNData *output, uint32_t nb, AVFilterContext *filter_ctx)
static void	free_detect_labels (DnnDetectContext *ctx)
static int	read_detect_label_file (AVFilterContext *context)
static int	check_output_nb (DnnDetectContext *ctx, DNNBackendType backend_type, int output_nb)
static av_cold int	dnn_detect_init (AVFilterContext *context)
static int	dnn_detect_flush_frame (AVFilterLink *outlink, int64_t pts, int64_t *out_pts)
static int	dnn_detect_activate (AVFilterContext *filter_ctx)
static av_cold void	dnn_detect_uninit (AVFilterContext *context)
static int	config_input (AVFilterLink *inlink)

Variables
static const AVOptionArrayDef	anchor_array_def = { .sep = '&' }
static const AVOption	dnn_detect_options []
static enum AVPixelFormat	pix_fmts []
static const AVFilterPad	dnn_detect_inputs []
const FFFilter	ff_vf_dnn_detect

Detailed Description

implementing an object detecting filter using deep learning networks.

Definition in file vf_dnn_detect.c.

Macro Definition Documentation

OFFSET

#define OFFSET

(

x

)

Value:

offsetof(DnnDetectContext, dnnctx.x)

Definition at line 64 of file vf_dnn_detect.c.

OFFSET2

#define OFFSET2

(

x

)

Value:

offsetof(DnnDetectContext, x)

Definition at line 65 of file vf_dnn_detect.c.

FLAGS

#define FLAGS   AV_OPT_FLAG_FILTERING_PARAM | AV_OPT_FLAG_VIDEO_PARAM

Definition at line 66 of file vf_dnn_detect.c.

Enumeration Type Documentation

DNNDetectionModelType

enum DNNDetectionModelType

Enumerator
DDMT_SSD
DDMT_YOLOV1V2
DDMT_YOLOV3
DDMT_YOLOV4

Definition at line 37 of file vf_dnn_detect.c.

Function Documentation

AVFILTER_DNN_DEFINE_CLASS()

AVFILTER_DNN_DEFINE_CLASS	(	dnn_detect	,
		DNN_TF|DNN_OV|	DNN_ONNX )

sigmoid()

static float sigmoid ( float x )

inlinestatic

Definition at line 94 of file vf_dnn_detect.c.

Referenced by dnn_detect_parse_yolo_output().

linear()

static float linear ( float x )

inlinestatic

Definition at line 98 of file vf_dnn_detect.c.

dnn_detect_get_label_id()

static int dnn_detect_get_label_id ( int nb_classes, int cell_size, float * label_data )

static

Definition at line 102 of file vf_dnn_detect.c.

Referenced by dnn_detect_parse_yolo_output().

dnn_detect_label_id_from_float()

static int dnn_detect_label_id_from_float ( float f )

static

Definition at line 115 of file vf_dnn_detect.c.

Referenced by dnn_detect_post_proc_ssd(), and dnn_detect_post_proc_tf().

dnn_detect_float_to_coord()

static int dnn_detect_float_to_coord ( double v )

static

Definition at line 122 of file vf_dnn_detect.c.

Referenced by dnn_detect_set_bbox_edges().

dnn_detect_confidence_num()

static int dnn_detect_confidence_num ( double conf )

static

Definition at line 133 of file vf_dnn_detect.c.

Referenced by dnn_detect_parse_yolo_output(), and dnn_detect_post_proc_ssd().

dnn_detect_set_bbox_edges()

static void dnn_detect_set_bbox_edges ( AVDetectionBBox * bbox, double x0, double y0, double x1, double y1, int frame_w, int frame_h )

static

Definition at line 145 of file vf_dnn_detect.c.

Referenced by dnn_detect_parse_yolo_output(), and dnn_detect_post_proc_ssd().

dnn_detect_IOU()

static float dnn_detect_IOU ( AVDetectionBBox * bbox1, AVDetectionBBox * bbox2 )

static

Definition at line 168 of file vf_dnn_detect.c.

Referenced by dnn_detect_fill_side_data().

dnn_detect_parse_yolo_output()

static int dnn_detect_parse_yolo_output ( AVFrame * frame, DNNData * output, int output_index, AVFilterContext * filter_ctx, int64_t * anchor_used )

static

Definition at line 187 of file vf_dnn_detect.c.

Referenced by dnn_detect_post_proc_yolo(), and dnn_detect_post_proc_yolov3().

dnn_detect_fill_side_data()

static int dnn_detect_fill_side_data ( AVFrame * frame, AVFilterContext * filter_ctx )

static

Definition at line 364 of file vf_dnn_detect.c.

Referenced by dnn_detect_post_proc_yolo(), and dnn_detect_post_proc_yolov3().

dnn_detect_post_proc_yolo()

static int dnn_detect_post_proc_yolo ( AVFrame * frame, DNNData * output, AVFilterContext * filter_ctx )

static

Definition at line 413 of file vf_dnn_detect.c.

Referenced by dnn_detect_post_proc_anchored().

dnn_detect_post_proc_yolov3()

static int dnn_detect_post_proc_yolov3 ( AVFrame * frame, DNNData * output, AVFilterContext * filter_ctx, int nb_outputs )

static

Definition at line 426 of file vf_dnn_detect.c.

Referenced by dnn_detect_post_proc_anchored().

dnn_detect_post_proc_ssd()

static int dnn_detect_post_proc_ssd ( AVFrame * frame, DNNData * output, int nb_outputs, AVFilterContext * filter_ctx )

static

Definition at line 442 of file vf_dnn_detect.c.

Referenced by dnn_detect_post_proc_anchored().

dnn_detect_post_proc_anchored()

static int dnn_detect_post_proc_anchored ( AVFrame * frame, DNNData * output, int nb_outputs, AVFilterContext * filter_ctx )

static

Definition at line 611 of file vf_dnn_detect.c.

Referenced by dnn_detect_post_proc().

dnn_detect_post_proc_tf()

static int dnn_detect_post_proc_tf ( AVFrame * frame, DNNData * output, AVFilterContext * filter_ctx )

static

Definition at line 645 of file vf_dnn_detect.c.

Referenced by dnn_detect_post_proc().

dnn_detect_post_proc()

static int dnn_detect_post_proc ( AVFrame * frame, DNNData * output, uint32_t nb, AVFilterContext * filter_ctx )

static

Definition at line 722 of file vf_dnn_detect.c.

Referenced by dnn_detect_init().

free_detect_labels()

static void free_detect_labels ( DnnDetectContext * ctx )

static

Definition at line 739 of file vf_dnn_detect.c.

Referenced by dnn_detect_uninit().

read_detect_label_file()

static int read_detect_label_file ( AVFilterContext * context )

static

Definition at line 748 of file vf_dnn_detect.c.

Referenced by dnn_detect_init().

check_output_nb()

static int check_output_nb ( DnnDetectContext * ctx, DNNBackendType backend_type, int output_nb )

static

Definition at line 806 of file vf_dnn_detect.c.

Referenced by dnn_detect_init().

dnn_detect_init()

static av_cold int dnn_detect_init ( AVFilterContext * context )

static

Definition at line 827 of file vf_dnn_detect.c.

dnn_detect_flush_frame()

static int dnn_detect_flush_frame ( AVFilterLink * outlink, int64_t pts, int64_t * out_pts )

static

Definition at line 871 of file vf_dnn_detect.c.

Referenced by dnn_detect_activate().

dnn_detect_activate()

static int dnn_detect_activate ( AVFilterContext * filter_ctx )

static

Definition at line 899 of file vf_dnn_detect.c.

dnn_detect_uninit()

static av_cold void dnn_detect_uninit ( AVFilterContext * context )

static

Definition at line 955 of file vf_dnn_detect.c.

config_input()

static int config_input ( AVFilterLink * inlink )

static

Definition at line 971 of file vf_dnn_detect.c.

Variable Documentation

anchor_array_def

const AVOptionArrayDef anchor_array_def = { .sep = '&' }

static

Definition at line 62 of file vf_dnn_detect.c.

dnn_detect_options

const AVOption dnn_detect_options[]

static

Initial value:

= {
    { "dnn_backend", "DNN backend",                OFFSET(backend_type),     AV_OPT_TYPE_INT,       { .i64 = DNN_OV },    INT_MIN, INT_MAX, FLAGS, .unit = "backend" },
 
 
 
 
 
 
 
 
 
    { "confidence",  "threshold of confidence",    OFFSET2(confidence),      AV_OPT_TYPE_FLOAT,     { .dbl = 0.5 },  0, 1, FLAGS},
    { "labels",      "path to labels file",        OFFSET2(labels_filename), AV_OPT_TYPE_STRING,    { .str = NULL }, 0, 0, FLAGS },
    { "model_type",  "DNN detection model type",   OFFSET2(model_type),      AV_OPT_TYPE_INT,       { .i64 = DDMT_SSD },    INT_MIN, INT_MAX, FLAGS, .unit = "model_type" },
        { "ssd",     "output shape [1, 1, N, 7]",  0,                        AV_OPT_TYPE_CONST,       { .i64 = DDMT_SSD },    0, 0, FLAGS, .unit = "model_type" },
        { "yolo",    "output shape [1, N*Cx*Cy*DetectionBox]",  0,           AV_OPT_TYPE_CONST,       { .i64 = DDMT_YOLOV1V2 },    0, 0, FLAGS, .unit = "model_type" },
        { "yolov3",  "outputs shape [1, N*D, Cx, Cy]",  0,                   AV_OPT_TYPE_CONST,       { .i64 = DDMT_YOLOV3 },      0, 0, FLAGS, .unit = "model_type" },
        { "yolov4",  "outputs shape [1, N*D, Cx, Cy]",  0,                   AV_OPT_TYPE_CONST,       { .i64 = DDMT_YOLOV4 },    0, 0, FLAGS, .unit = "model_type" },
    { "cell_w",      "cell width",                 OFFSET2(cell_w),          AV_OPT_TYPE_INT,       { .i64 = 0 },    0, INTMAX_MAX, FLAGS },
    { "cell_h",      "cell height",                OFFSET2(cell_h),          AV_OPT_TYPE_INT,       { .i64 = 0 },    0, INTMAX_MAX, FLAGS },
    { "nb_classes",  "The number of class",        OFFSET2(nb_classes),      AV_OPT_TYPE_INT,       { .i64 = 0 },    0, INTMAX_MAX, FLAGS },
    { "anchors",     "anchors, split by '&'",    OFFSET2(anchors),         AV_OPT_TYPE_FLOAT | AV_OPT_TYPE_FLAG_ARRAY,    { .arr = &anchor_array_def }, FLT_MIN, FLT_MAX, FLAGS },
    { NULL }
}

Definition at line 67 of file vf_dnn_detect.c.

pix_fmts

enum AVPixelFormat pix_fmts[]

static

Initial value:

= {
    AV_PIX_FMT_RGB24, AV_PIX_FMT_BGR24,
    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
}

Definition at line 862 of file vf_dnn_detect.c.

dnn_detect_inputs

const AVFilterPad dnn_detect_inputs[]

static

Initial value:

= {
    {
        .name         = "default",
        .type         = AVMEDIA_TYPE_VIDEO,
        .config_props = config_input,
    },
}

Definition at line 993 of file vf_dnn_detect.c.

ff_vf_dnn_detect

const FFFilter ff_vf_dnn_detect

Initial value:

= {
    .p.name        = "dnn_detect",
    .p.description = NULL_IF_CONFIG_SMALL("Apply DNN detect filter to the input."),
    .p.priv_class  = &dnn_detect_class,
    .priv_size     = sizeof(DnnDetectContext),
    .preinit       = ff_dnn_filter_init_child_class,
    .init          = dnn_detect_init,
    .uninit        = dnn_detect_uninit,
    FILTER_INPUTS(dnn_detect_inputs),
    FILTER_OUTPUTS(ff_video_default_filterpad),
    FILTER_PIXFMTS_ARRAY(pix_fmts),
    .activate      = dnn_detect_activate,
}

Definition at line 1001 of file vf_dnn_detect.c.