Nabrio Help
Nabrio Help

Getting Started

Nara Overview

Understanding Nara

Using Nara

Components

Common Process InputsCombine ArraysFilter Array OutputGet Array ItemGet Array Object ItemGet Last Run ArrayGet Object ItemJSON String to ObjectBasic MathsGet Last Run Upstream OutputsMoving Number CalculationNumber RoundingNumber to StringBased64 string to ImageString to NumberTemplate StringISO Time DeltaXML String to ObjectAuto Label Object DetectionBrightnessCircle Detection1D/2D Code ReaderColor DetectionColor ThresholdContrastMultiple CropsCropDetected Zones CountersFacial Expression DetectionGender and Age DetectionDetected Objects to OCRDetected Objects TransformPeople DetectionDraw Detected ObjectsFace DetectionFace RecognitionFeature MatchingFire & Smoke DetectionFlipGeneral Object DetectionImage ClassificationImage SimilarityKey Points DetectionNumber Plate ReaderObject DetectionOCRPose EstimationResizeRotateSaturationInstance SegmentationWatershed SegmentPolygon DetectionTracker Line CountTracker Zones CountTracker HeatmapTrackerWhite Balance
Widgets

Miscellaneous

Nomenclature
Troubleshooting
Notice and DisclaimerEULA
NodesProcess

Facial Expression Detection

Slot Usage: 1

Overview

Facial Expression Detection node classifies facial expressions for detected faces and returns an overlay, a summed expression score, and enriched face attributes.

Connect faces from Face Detection. OpenCV Face Expression Recognition does not include contempt; UniFace AffecNet8 supports the full expression set.

Input

Input Image

image required

The image frame to analyze. Connect this to a camera or upstream image output.

Detected Faces

array required

List of faces detected in the frame. Each object must include at least a Bounding Box (bbox: [x, y, width, height]). Label, Confidence, and Extra Data use defaults when missing.

Facial Expression Model

string required advanced

Model used for facial expression classification.

Values:

  • ONUF_AFFECNET8 (default) — UniFace: AffecNet8
  • ONCV_FACE_EXPR_RECOG — OpenCV: Face Expression Recognition

Confidence threshold

number required advanced

Minimum confidence score to keep a prediction. See Confidence Threshold for tuning guidance.

Range: 0.0 to 1.0
Default: 0.5

Minimum Face Size

array required advanced

Minimum [width, height] of a face box for expression detection.

Default: [25, 25]

Add attributes to person

boolean advanced

When enabled, also adds expression attributes to the parent person object that contains the face.

Default: true

Use TensorRT if available

boolean required advanced

Enable TensorRT inference backend when available. Can improve throughput on NVIDIA GPUs.

Default: false

Expression Score Map

object advanced

Numeric score mapping used when summing expression scores. Defaults:

  • Neutral: 0
  • Happy: 1
  • Sad: -1
  • Surprised: 2
  • Fear: -3
  • Disgusted: -2
  • Angry: -3
  • Contempt: -2

Overlay results

boolean required advanced

Whether to draw results on the output frame. See Overlay Results.

Default: true

Output

Overlay Image

image

Frame with expression overlays when overlay is enabled.

Summed Expression Scores

number

Sum of mapped expression scores for faces in the current frame.

Detected Faces

array

Enriched face detections. Each object includes the original fields plus:

  • Facial Expression — neutral, happy, sad, surprised, fear, disgusted, angry, contempt, or unknown
  • Expression Score

Detected Zones Counters

Previous Page

Gender and Age Detection

Next Page

On this page

OverviewInputInput ImageDetected FacesFacial Expression ModelConfidence thresholdMinimum Face SizeAdd attributes to personUse TensorRT if availableExpression Score MapOverlay resultsOutputOverlay ImageSummed Expression ScoresDetected Faces