Generates a list of detections based on location and confidence predictions by doing non maximum suppression. More...

#include <detection_output.hpp>

Inheritance diagram for cldnn::detection_output:

Collaboration diagram for cldnn::detection_output:

[legend]

Public Types
typedef cldnn_detection_output_desc	dto

Public Member Functions
	detection_output (const primitive_id &id, const primitive_id &input_location, const primitive_id &input_confidence, const primitive_id &input_prior_box, const uint32_t num_classes, const uint32_t keep_top_k, const bool share_location=true, const int background_label_id=0, const float nms_threshold=0.3, const int top_k=-1, const float eta=1.f, const prior_box_code_type code_type=prior_box_code_type::corner, const bool variance_encoded_in_target=false, const float confidence_threshold=-std::numeric_limits< float >::max(), const padding &output_padding=padding())
	Constructs pooling primitive. More...

	detection_output (const dto *dto)
	Constructs a copy from C API cldnn_detection_output_desc.

Public Member Functions inherited from cldnn::primitive_base< detection_output, cldnn_detection_output_desc >
const cldnn_primitive_desc *	get_dto () const override
	Returns pointer to a C API primitive descriptor casted to cldnn_primitive_desc.

Public Member Functions inherited from cldnn::primitive
	primitive (const primitive_type_id &type, const primitive_id &id, const std::vector< primitive_id > &input, const padding &output_padding=padding())

	primitive (const cldnn_primitive_desc *dto)
	Constructs a copy from basic C API cldnn_primitive_desc.

std::vector< std::reference_wrapper< primitive_id > >	dependecies ()
	Returns references to all primitive ids on which this primitive depends - inputs, weights, biases, etc.

std::vector< primitive_id >	dependecies () const
	Returns copy of all primitive ids on which this primitive depends - inputs, weights, biases, etc.

	operator primitive_id () const
	Implicit conversion to primiitive id.

Static Public Member Functions
static primitive_type_id	type_id ()

Public Attributes
const uint32_t	num_classes
	Number of classes to be predicted.

const int	keep_top_k
	Number of total bounding boxes to be kept per image after NMS step.

const bool	share_location
	If true, bounding box are shared among different classes.

const int	background_label_id
	Background label id (-1 if there is no background class).

const float	nms_threshold
	Threshold for NMS step.

const int	top_k
	Maximum number of results to be kept in NMS.

const float	eta
	Used for adaptive NMS.

const prior_box_code_type	code_type
	Type of coding method for bounding box.

const bool	variance_encoded_in_target
	If true, variance is encoded in target; otherwise we need to adjust the predicted offset accordingly.

const float	confidence_threshold
	Only keep detections with confidences larger than this threshold.

Public Attributes inherited from cldnn::primitive
const primitive_type_id	type
	Primitive's type id.

const primitive_id	id
	Primitive's id.

fixed_size_vector_ref	input
	List of ids of input primitives.

padding	output_padding
	Requested output padding.

Protected Member Functions
void	update_dto (dto &dto) const override

Protected Member Functions inherited from cldnn::primitive_base< detection_output, cldnn_detection_output_desc >
	primitive_base (const primitive_id &id, const std::vector< primitive_id > &input, const padding &output_padding=padding())

	primitive_base (const cldnn_detection_output_desc *dto)

Protected Member Functions inherited from cldnn::primitive
virtual std::vector< std::reference_wrapper< const primitive_id > >	get_dependencies () const

Additional Inherited Members
Protected Attributes inherited from cldnn::primitive
primitive_id_arr	_input

Detailed Description

Generates a list of detections based on location and confidence predictions by doing non maximum suppression.

Each row is a 7 dimension vector, which stores: [image_id, label, confidence, xmin, ymin, xmax, ymax]. If number of detections per image is lower than keep_top_k, will write dummy results at the end with image_id=-1.

Definition at line 43 of file detection_output.hpp.

Constructor & Destructor Documentation

◆ detection_output()

cldnn::detection_output::detection_output	(	const primitive_id &	id,
		const primitive_id &	input_location,
		const primitive_id &	input_confidence,
		const primitive_id &	input_prior_box,
		const uint32_t	num_classes,
		const uint32_t	keep_top_k,
		const bool	share_location = `true`,
		const int	background_label_id = `0`,
		const float	nms_threshold = `0.3`,
		const int	top_k = `-1`,
		const float	eta = `1.f`,
		const prior_box_code_type	code_type = `prior_box_code_type::corner`,
		const bool	variance_encoded_in_target = `false`,
		const float	confidence_threshold = `-std::numeric_limits<float>::max()`,
		const padding &	output_padding = `padding()`
	)

inline

Constructs pooling primitive.

Parameters

id	This primitive id.
input_location	Input location primitive id.
input_confidence	Input confidence primitive id.
input_prior_box	Input prior-box primitive id.
num_classes	Number of classes to be predicted.
keep_top_k	Number of total bounding boxes to be kept per image after NMS step.
share_location	If true bounding box are shared among different classes.
background_label_id	Background label id (-1 if there is no background class).
nms_threshold	Threshold for NMS step.
top_k	Maximum number of results to be kept in NMS.
eta	Used for adaptive NMS.
code_type	Type of coding method for bounding box.
variance_encoded_in_target	If true, variance is encoded in target; otherwise we need to adjust the predicted offset accordingly.
confidence_threshold	Only keep detections with confidences larger than this threshold.

Definition at line 62 of file detection_output.hpp.

The documentation for this struct was generated from the following file:

C:/ML/OpenSource/clDNN/api/CPP/detection_output.hpp

Public Types

Public Member Functions

Static Public Member Functions

Public Attributes

Protected Member Functions

Additional Inherited Members

Detailed Description

Constructor & Destructor Documentation

◆ detection_output()