ObjectDetectionParameters Class Reference

Sets the object detection parameters. More...


def __cinit__ (self, image_sync=True, enable_tracking=True, enable_mask_output=True, detection_model=DETECTION_MODEL.MULTI_CLASS_BOX, enable_body_fitting=False, max_range=-1.0, batch_trajectories_parameters=BatchParameters(), body_format=BODY_FORMAT.POSE_18, filtering_mode=OBJECT_FILTERING_MODE.NMS3D, prediction_timeout_s=0.2, allow_reduced_precision_inference=False)
 Constructor. More...
def image_sync (self)
 Defines if the object detection is synchronized to the image or runs in a separate thread.
def enable_tracking (self)
 Defines if the object detection will track objects across images flow.
def enable_mask_output (self)
 Defines if the mask object will be computed.
def detection_model (self)
 Enable human pose estimation with skeleton keypoints output.
def body_format (self)
 Defines the body format output by the SDK when retrieve_objects is called. More...
def enable_body_fitting (self)
 Defines if the body fitting will be applied.
def max_range (self)
 Defines an upper depth range for detections
Defined in InitParameters.coordinate_units
Default value is set to InitParameters.depth_maximum_distance (can not be higher)
def batch_parameters (self)
 Batching system (introduced in 3.5) performs short-term re-identification with deep learning and trajectories filtering. More...
def filtering_mode (self)
 Filtering mode for MULTI_CLASS_BOX and Custom objects tracking.
def prediction_timeout_s (self)
 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.
def allow_reduced_precision_inference (self)
 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.

Detailed Description

Sets the object detection parameters.

The default constructor sets all parameters to their default settings.

Parameters can be user adjusted.


◆ __cinit__()

def __cinit__ (   self,
  image_sync = True,
  enable_tracking = True,
  enable_mask_output = True,
  enable_body_fitting = False,
  max_range = -1.0,
  batch_trajectories_parameters = BatchParameters(),
  body_format = BODY_FORMAT.POSE_18,
  filtering_mode = OBJECT_FILTERING_MODE.NMS3D,
  prediction_timeout_s = 0.2,
  allow_reduced_precision_inference = False 


Calling the constructor without any parameter will set them to their default values.

image_sync: sets image_sync. Default: True
enable_tracking: sets enable_tracking. Default: True
enable_mask_output: sets enable_mask_output. Default: True
enable_body_fitting: sets enable_body_fitting. Default: False
max_range: sets max_range. Default: -1.0 (set to InitParameters.depth_maximum_distance)
batch_trajectories_parameters: sets batch_parameters. Default: see BatchParameters default constructor
body_format: sets body_format. Default: sl.BODY_FORMAT.POSE_18

◆ body_format()

def body_format (   self)

Defines the body format output by the SDK when retrieve_objects is called.

if set to sl.BODY_FORMAT.POSE_32, the ZED SDK will automatically enable the fitting (cf. enable_body_fitting).

◆ batch_parameters()

def batch_parameters (   self)

Batching system (introduced in 3.5) performs short-term re-identification with deep learning and trajectories filtering.

BatchParameters.enable needs to be set to True to use this feature (by default, it is disabled)