FacesToolConfig

Field

Type

Description

confidence_threshold

double

score = sqrt(cls*obj); SDK defaults to 0.3

aggregation

FacesAggregation

sample_n

int32

bucket_secs

double

for BUCKETED; SDK defaults to 1.0

infer_every_ms

double

min media-time between inferences; 0 = every frame

match_threshold

double

Face recognition (SFace) — enabled by a non-empty gallery. Enrolment images are raw rgb24 with one prominent subject (the largest confident face is enrolled); a bad image fails configuration.

0 = default 0.4 (OpenCV’s canonical 0.363 + margin)

gallery

FaceRef

model

string

"" or "yunet" (today’s model; named for forward compatibility)