Tracking

Algorithms for tracking the objects generated from the clustering

Table 1. Available fields in blickfeld.percept_pipeline.config.Tracking
Field Type Default Unit

disable

bool

-

-

Disable tracking. On by default.

sensitivity

float

-

-

Higher sensitivity will make the tracker confirm the object faster

Reasonable default: 0.85

acceleration_noise

float

-

-

The acceptance range of accelerations for the object to be tracked

Reasonable default: 0.3

survival_probability

float

-

-

Higher probability will allow the object to be followed longer in the absence of observations

Reasonable default: 1

path_resolution

float

-

-

The distance between saved object positions in the track path history. The path history is stored for all currently tracked objects. The path resolution specifies the discretization of the stored path. A higher value results in greater spacing between recorded positions.

Default value: 1 meter

advanced

Advanced

-

-

Advanced tracking settings

NOTE: If set, configuration will override parameters in higher level tracking configuration.

Advanced

Advanced tracking configuration settings

Table 2. Available fields in blickfeld.percept_pipeline.config.Tracking.Advanced
Field Type Default Unit

sensitivity

float

-

-

Defines the speed and which the tracker confirms the object. Higher values initiate tracking of a detected object even when the uncertainty about its existence is high. Lower values reduce the sensitivity, meaning that tracking will only begin when an object is clearly detected in multiple consecutive frames.

default: 0.85, range: [0.01 …​ 1.0]

acceleration_noise

float

-

m^2 / s^4

Defines the amount of acceleration noise that can happen in any random direction. When tracking fails to properly follow fast-moving objects, increasing this value expands the search space for track association.

default: 0.3, range: [0.3 …​ 10]

observation_noise

float

-

-

Describes the expected deviation squared (variance) of the detection from the true center of the object

default: 0.3

minimum_association_probability_for_detection

float

-

-

Defines the threshold below which a detection is ignored from being associated into the track

default 0.2, range: [0.01 …​ 0.9999]

gate_probability

float

-

-

Defines how often do true detections happen within the validation gate

default: 0.9995, range: [0.01 …​ 0.9999]

detection_probability

float

-

-

Defines the probability expected to detect a true object in the scene. Influences the algorithm in three ways: contributes to the estimation of clutter measurements, the possibility of associating a missed detection to an existing track and the final probability of missed detection.

default: 0.88, range: [0.01 …​ 0.9999]

survival_probability

float

-

-

Defines how many frames are used to keep objects without detections alive. High values will keep the track alive for an extended period. This can be useful to pick up the same objects' ids again after the previously tracked object has been hidden for some frames. Lower values cause the system to discard non-visible tracks faster.

default: 0.995, range: [0.01 …​ 0.9999]

confirmed_existence_probability

float

-

-

Defines how quickly tracks are confirmed for detections

default: 0.75, range: [0.01 …​ 0.9999]

minimum_existence_probability

float

-

-

Describes the existence probability at which a track gets deleted

default: 0.01, range: [0.01 …​ 0.9999]

initial_position_uncertainty

float

-

m^2

Determines the expected uncertainty in the position. This influences the search radius for associating the second detection to this object.

default: 1.0

initial_velocity_uncertainty

float

-

m^2 / s^2

Determines the expected uncertainty in the velocity. Needs to be increased if the object velocity, while entering the scene, is high, and can be decreased if the entering speeds are low. A benefit of reducing this parameter is that the distance between two objects can be smaller for the tracker to still be able to resolve them.

default: 10