ObjectDetectionParameters Class Reference

Sets the object detection parameters. More...

Functions
	ObjectDetectionParameters (bool image_sync_=true, bool enable_tracking_=true, bool enable_mask_output_=false, DETECTION_MODEL detection_model=DETECTION_MODEL::MULTI_CLASS_BOX, bool enable_body_fitting_=false, float max_range_=-1.f, BatchParameters batch_trajectories_parameters=BatchParameters(), BODY_FORMAT body_format_=BODY_FORMAT::POSE_18, OBJECT_FILTERING_MODE filtering_mode_=OBJECT_FILTERING_MODE::NMS3D, float prediction_timeout_s=0.2f, bool allow_reduced_precision_inference=false)
	Default constructor. Set all parameters to their default values. More...
bool	operator== (const ObjectDetectionParameters &param1) const
bool	operator!= (const ObjectDetectionParameters &param1) const

Attributes
bool	image_sync = true
	Defines if the object detection is synchronized to the image or runs in a separate thread. If set to true, the detection is run for every grab, otherwise, the thread runs at its own speed, which can lead to new detection once in a while. More...
bool	enable_tracking = true
	Defines if the object detection will track objects across images flow. More...
bool	enable_mask_output = false
	Defines if the mask object will be computed. More...
DETECTION_MODEL	detection_model = DETECTION_MODEL::MULTI_CLASS_BOX
	Enable human pose estimation with skeleton keypoints output. More...
bool	enable_body_fitting = false
	Defines if the body fitting will be applied. More...
BODY_FORMAT	body_format = BODY_FORMAT::POSE_18
	Defines the body format outputed by the sdk when retrieveObjects is called. More...
float	max_range = -1.f
	Defines a upper depth range for detections. Defined in UNIT set at sl::Camera::open. Default value is set to sl::Initparameters::depth_maximum_distance (can not be higher). More...
BatchParameters	batch_parameters
	Batching system parameters. Batching system (introduced in 3.5) performs short-term re-identification with deep learning and trajectories filtering. BatchParameters::enable need to be true to use this feature (by default disabled) More...
OBJECT_FILTERING_MODE	filtering_mode
	Defines the filtering mode that should be applied to raw detections. Default value is set to OBJECT_FILTERING_MODE::NMS_3D to keep the same behavior as previous SDK version. More...
float	prediction_timeout_s
	When an object is not detected anymore, the SDK will predict its positions during a short period of time before its state switched to SEARCHING. It prevents the jittering of the object state when there is a short misdetection. The user can define its own prediction time duration. More...
bool	allow_reduced_precision_inference
	Allow inference to run at a lower precision to improve runtime and memory usage, it might increase the initial optimization time and could include downloading calibration data or calibration cache and slightly reduce the accuracy. More...

Detailed Description

Sets the object detection parameters.

The default constructor sets all parameters to their default settings.

Note

Parameters can be user adjusted.

Constructor and Destructor

ObjectDetectionParameters()

ObjectDetectionParameters	(	bool	image_sync_ = true,
		bool	enable_tracking_ = true,
		bool	enable_mask_output_ = false,
		DETECTION_MODEL	detection_model = DETECTION_MODEL::MULTI_CLASS_BOX,
		bool	enable_body_fitting_ = false,
		float	max_range_ = -1.f,
		BatchParameters	batch_trajectories_parameters = BatchParameters(),
		BODY_FORMAT	body_format_ = BODY_FORMAT::POSE_18,
		OBJECT_FILTERING_MODE	filtering_mode_ = OBJECT_FILTERING_MODE::NMS3D,
		float	prediction_timeout_s = 0.2f,
		bool	allow_reduced_precision_inference = false
	)

Default constructor. Set all parameters to their default values.

Functions

operator==()

bool operator==

(

const ObjectDetectionParameters &

param1

)

const

Comparison operator ==

Parameters

ObjectDetectionParameters

to compare

Returns

true if the two struct are identical

operator!=()

bool operator!=

(

const ObjectDetectionParameters &

param1

)

const

Comparison operator !=

Parameters

ObjectDetectionParameters

to compare

Returns

true if the two struct are different

Variables

image_sync

bool image_sync = true

Defines if the object detection is synchronized to the image or runs in a separate thread. If set to true, the detection is run for every grab, otherwise, the thread runs at its own speed, which can lead to new detection once in a while.

enable_tracking

bool enable_tracking = true

Defines if the object detection will track objects across images flow.

enable_mask_output

bool enable_mask_output = false

Defines if the mask object will be computed.

detection_model

DETECTION_MODEL detection_model = DETECTION_MODEL::MULTI_CLASS_BOX

Enable human pose estimation with skeleton keypoints output.

enable_body_fitting

bool enable_body_fitting = false

Defines if the body fitting will be applied.

body_format

BODY_FORMAT body_format = BODY_FORMAT::POSE_18

Defines the body format outputed by the sdk when retrieveObjects is called.

Warning

if BODY_FORMAT::POSE_34, the ZED SDK will automatically enable the fitting enable_body_fitting

max_range

float max_range = -1.f

Defines a upper depth range for detections.
Defined in UNIT set at sl::Camera::open.
Default value is set to sl::Initparameters::depth_maximum_distance (can not be higher).

batch_parameters

BatchParameters batch_parameters

Batching system parameters. Batching system (introduced in 3.5) performs short-term re-identification with deep learning and trajectories filtering.
BatchParameters::enable need to be true to use this feature (by default disabled)

filtering_mode

OBJECT_FILTERING_MODE filtering_mode

Defines the filtering mode that should be applied to raw detections.
Default value is set to OBJECT_FILTERING_MODE::NMS_3D to keep the same behavior as previous SDK version.

Note

: This parameter is only used in Detection Model MULTI_CLASS_BOX_XXXX and CUSTOM_BOX_OBJECTS

: For Custom Object mode, it is recommended to use OBJECT_FILTERING_MODE::NMS_3D_PER_CLASS or OBJECT_FILTERING_MODE::NONE. In this case, you might need to add your own NMS filter before ingesting the boxes into the object detection module.

prediction_timeout_s

float prediction_timeout_s

When an object is not detected anymore, the SDK will predict its positions during a short period of time before its state switched to SEARCHING.
It prevents the jittering of the object state when there is a short misdetection. The user can define its own prediction time duration.

Note

During this time, the object will have OK state even if it is not detected.

the duration is expressed in seconds

Warning

the prediction_timeout_s will be clamped to 1 second as the prediction is getting worst with time.

set this parameter to 0 to disable SDK predictions

allow_reduced_precision_inference

bool allow_reduced_precision_inference

Allow inference to run at a lower precision to improve runtime and memory usage, it might increase the initial optimization time and could include downloading calibration data or calibration cache and slightly reduce the accuracy.

Note

The fp16 is automatically enabled if the GPU is compatible and provides a speed up of almost x2 and reduce memory usage by almost half, no precision loss.

This setting allow int8 precision which can speed up by another x2 factor (compared to fp16, or x4 compared to fp32) and half the fp16 memory usage, however some accuracy can be lost. The accuracy loss should not exceed 1-2% on the compatible models. The current compatible models are all HUMAN_BODY_XXXX