FFmpeg: libavfilter/dnn/dnn_backend_common.h File Reference

FFmpeg

[フレーム]

Data Structures | Macros | Functions

dnn_backend_common.h File Reference

#include "queue.h"
#include "../dnn_interface.h"
#include "libavutil/thread.h"

Go to the source code of this file.

Data Structures

struct TaskItem

struct LastLevelTaskItem

struct DNNAsyncExecModule

Common Async Execution Mechanism for the DNN Backends. More...

Macros

#define DNN_BACKEND_COMMON_OPTIONS

Functions

int ff_check_exec_params (void *ctx, DNNBackendType backend, DNNFunctionType func_type, DNNExecBaseParams *exec_params)

int ff_dnn_fill_task (TaskItem *task, DNNExecBaseParams *exec_params, void *backend_model, int async, int do_ioproc)

Fill the Task for Backend Execution. More...

int ff_dnn_async_module_cleanup (DNNAsyncExecModule *async_module)

Join the Async Execution thread and set module pointers to NULL. More...

int ff_dnn_start_inference_async (void *ctx, DNNAsyncExecModule *async_module)

Start asynchronous inference routine for the TensorFlow model on a detached thread. More...

DNNAsyncStatusType ff_dnn_get_result_common (Queue *task_queue, AVFrame **in, AVFrame **out)

Extract input and output frame from the Task Queue after asynchronous inference. More...

int ff_dnn_fill_gettingoutput_task (TaskItem *task, DNNExecBaseParams *exec_params, void *backend_model, int input_height, int input_width, void *ctx)

Allocate input and output frames and fill the Task with execution parameters. More...

Detailed Description

DNN common functions different backends.

Definition in file dnn_backend_common.h.

Macro Definition Documentation

◆ DNN_BACKEND_COMMON_OPTIONS

#define DNN_BACKEND_COMMON_OPTIONS

Value:

{ "nireq", "number of request", OFFSET(options.nireq), AV_OPT_TYPE_INT, { .i64 = 0 }, 0, INT_MAX, FLAGS }, \

{ "async", "use DNN async inference", OFFSET(options.async), AV_OPT_TYPE_BOOL, { .i64 = 1 }, 0, 1, FLAGS },

Definition at line 31 of file dnn_backend_common.h.

Function Documentation

◆ ff_check_exec_params()

int ff_check_exec_params ( void * ctx,

DNNBackendType backend,

DNNFunctionType func_type,

DNNExecBaseParams * exec_params

)

Definition at line 29 of file dnn_backend_common.c.

Referenced by ff_dnn_execute_model_native(), ff_dnn_execute_model_ov(), and ff_dnn_execute_model_tf().

◆ ff_dnn_fill_task()

int ff_dnn_fill_task ( TaskItem * task,

DNNExecBaseParams * exec_params,

void * backend_model,

int async,

int do_ioproc

)

Fill the Task for Backend Execution.

It should be called after checking execution parameters using ff_check_exec_params.

Parameters: task pointer to the allocated task

exec_param pointer to execution parameters

backend_model void pointer to the backend model

async flag for async execution. Must be 0 or 1

do_ioproc flag for IO processing. Must be 0 or 1

Returns: 0 if successful or error code otherwise.

Definition at line 56 of file dnn_backend_common.c.

Referenced by ff_dnn_execute_model_native(), ff_dnn_execute_model_ov(), ff_dnn_execute_model_tf(), and ff_dnn_fill_gettingoutput_task().

◆ ff_dnn_async_module_cleanup()

int ff_dnn_async_module_cleanup ( DNNAsyncExecModule * async_module )

Join the Async Execution thread and set module pointers to NULL.

Parameters: async_module pointer to DNNAsyncExecModule module

Returns: 0 if successful or error code otherwise.

Definition at line 92 of file dnn_backend_common.c.

Referenced by destroy_request_item().

◆ ff_dnn_start_inference_async()

int ff_dnn_start_inference_async ( void * ctx,

DNNAsyncExecModule * async_module

)

Start asynchronous inference routine for the TensorFlow model on a detached thread.

It calls the completion callback after the inference completes. Completion callback and inference function must be set before calling this function.

If POSIX threads aren't supported, the execution rolls back to synchronous mode, calling completion callback after inference.

Parameters: ctx pointer to the backend context

async_module pointer to DNNAsyncExecModule module

Returns: 0 on the start of async inference or error code otherwise.

Definition at line 111 of file dnn_backend_common.c.

Referenced by execute_model_tf(), and ff_dnn_flush_tf().

◆ ff_dnn_get_result_common()

DNNAsyncStatusType ff_dnn_get_result_common ( Queue * task_queue,

AVFrame ** in,

AVFrame ** out

)

Extract input and output frame from the Task Queue after asynchronous inference.

Parameters: task_queue pointer to the task queue of the backend

in double pointer to the input frame

out double pointer to the output frame

Return values: DAST_EMPTY_QUEUE if task queue is empty

DAST_NOT_READY if inference not completed yet.

DAST_SUCCESS if result successfully extracted

Definition at line 142 of file dnn_backend_common.c.

Referenced by ff_dnn_get_result_native(), ff_dnn_get_result_ov(), and ff_dnn_get_result_tf().

◆ ff_dnn_fill_gettingoutput_task()

int ff_dnn_fill_gettingoutput_task ( TaskItem * task,

DNNExecBaseParams * exec_params,

void * backend_model,

int input_height,

int input_width,

void * ctx

)

Allocate input and output frames and fill the Task with execution parameters.

Parameters: task pointer to the allocated task

exec_params pointer to execution parameters

backend_model void pointer to the backend model

input_height height of input frame

input_width width of input frame

ctx pointer to the backend context

Returns: 0 if successful or error code otherwise.

Definition at line 162 of file dnn_backend_common.c.

Referenced by get_output_native(), get_output_ov(), and get_output_tf().

FLAGS

#define FLAGS

Definition: cmdutils.c:509

options

const OptionDef options[]

OFFSET

it s the only field you need to keep assuming you have a context There is some magic you don t need to care about around this just let it vf default minimum maximum flags name is the option keep it simple and lowercase description are in without and describe what they for example set the foo of the bar offset is the offset of the field in your see the OFFSET() macro

AV_OPT_TYPE_INT

@ AV_OPT_TYPE_INT

Definition: opt.h:225

AV_OPT_TYPE_BOOL

@ AV_OPT_TYPE_BOOL

Definition: opt.h:244

Generated on Wed Aug 24 2022 21:42:39 for FFmpeg by doxygen 1.8.17