Input
Executionexec_in
Initiate Execution
Associate Entities Node
AI/ML/Tracking
Gives local tracks from one or more cameras a global identity by comparing appearance embeddings. A track is collected for a few observations before it creates a new entity; later tracks that look like a known entity are matched to it, even on another camera. Every board and user of this app using the same task id shares the identities. They live in this process's memory only: after an hour without calls or a restart the task starts over and entity ids begin again at 1. An app can have at most 16 active tasks (64 per process); an idle task is released after an hour.
tracking_associate_entitiesonnx
Ratings
Scores range from 0 to 10. Higher values mean more impact, exposure, or operational weight.
SecurityAttack surface and exposure impact.
7/10High
PrivacyPotential sensitivity of processed data.
4/10Medium
PerformanceRuntime or resource pressure.
8/10High
GovernancePolicy, audit, or compliance impact.
4/10Medium
ReliabilityOperational stability considerations.
7/10High
CostExternal or compute cost impact.
10/10High
Input Pins
7Observations
Struct Arrayobservations
Observations with appearance embeddings, e.g. from Extract Appearance. Observations with a track id accumulate evidence per track; ones without are only matched to known entities, and lost tracks only report the entity they already have. Timestamps are Unix milliseconds from one clock shared by all cameras of the task; a missing timestamp means now, and one further ahead of this machine's clock than a minute (or half the entity lifetime, if shorter) is clamped.
AppearanceObservationAppearanceObservation15 fieldsDefault []
x1number:floatrequired
y1number:floatrequired
x2number:floatrequired
y2number:floatrequired
scorenumber:floatrequired
class_idxinteger:int32required
class_namestring | null
embeddingArray<number:float>
L2-normalized appearance embedding; empty when none was extracted
itemsnumber:floatarray item
camera_idstring
Camera the observation came from
session_idstring
Stream session of that camera
tracker_idstring
Tracker instance that issued `track_id`; track ids are only unique per tracker instance
track_idinteger | null
Local track id from Track Detections, if the observation belongs to a track
stateTrackState
State of the source track; plain detections are `tracked`
variant 1constvariant
Matched to a detection in the latest frame
variant 2constvariant
Not matched recently; kept alive for re-identification until it expires
timestamp_msinteger:int64
Frame capture time in Unix milliseconds
detection_indexinteger | null
Index of the source element in the producing node's input array
Task Id
Stringtask_id
Identities are shared by every board, user and camera of this app using the same task id
Default default
Min Similarity
Floatmin_similarity
Lowest cosine similarity for matching an observation to an entity
Default 0.4
Ambiguity Margin
Floatambiguity_margin
A match is left unresolved when the next best entity is at most this much less similar
Default 0.05
Min Observations
Integermin_observations
Observations a new track needs before it is matched or creates an entity (at least 1)
Default 3
Entity Lifetime (ms)
Integerentity_ttl_ms
Entities, track bindings and pending tracks unseen for this long, measured by observation timestamps, are forgotten
Default 600000
Output Pins
4Output
Executionexec_out
Done with the Execution
Associations
Struct Arrayassociations
Observations with their global entity id, in input order
EntityAssociationEntityAssociation16 fieldsDefault []
x1number:floatrequired
y1number:floatrequired
x2number:floatrequired
y2number:floatrequired
scorenumber:floatrequired
class_idxinteger:int32required
class_namestring | null
camera_idstring
session_idstring
tracker_idstring
track_idinteger | null
timestamp_msinteger:int64
detection_indexinteger | null
entity_idinteger:uint64required
Global entity id, unique within the association task
similaritynumber:floatrequired
Cosine similarity to the entity's prototype: of this observation when `matched` or `created`, of the track's last embedded observation when `tracked`
statusAssociationStatusrequired
variant 1constvariant
The local track was already bound to this entity
variant 2constvariant
The local track was newly matched to an existing entity
variant 3constvariant
No existing entity matched; a new one was created
Unresolved
Struct Arrayunresolved
Observations without an entity yet, with the reason and the closest entities
UnresolvedObservationUnresolvedObservation16 fieldsDefault []
x1number:floatrequired
y1number:floatrequired
x2number:floatrequired
y2number:floatrequired
scorenumber:floatrequired
class_idxinteger:int32required
class_namestring | null
camera_idstring
session_idstring
tracker_idstring
track_idinteger | null
timestamp_msinteger:int64
detection_indexinteger | null
reasonUnresolvedReasonrequired
variant 1constvariant
The local track has not been observed often enough to create an entity
variant 2constvariant
Two or more entities match within the ambiguity margin
variant 3constvariant
The observation carries no usable embedding
variant 4constvariant
No track id, so no evidence can accumulate and no entity is created
variant 5constvariant
The track is lost: its box is a prediction, so it neither matches nor creates an entity
observationsinteger:uint32required
Observations accumulated for this local track so far
candidatesArray<EntityCandidate>required
Closest entities, most similar first
itemsEntityCandidatearray item
entity_idinteger:uint64required
similaritynumber:floatrequired
Entity Count
Integerentity_count
Number of entities currently known to this task
Node Info
- Internal name
- tracking_associate_entities
- Category
- AI/ML/Tracking
- Version
- 1