The Face Detection extractor identifies human faces in images and videos, extracts facial features, and optionally matches them against known identities.

​
Overview

The Face Detection feature extractor analyzes visual content to locate human faces, extract facial embeddings, and optionally identify individuals based on pre-enrolled face data. It supports applications such as identity verification, person tracking, and face-based content search.

​
Required Inputs

ParameterTypeRequiredDefaultDescription
media_urlstringYes-URL pointing to image or video file. Supported formats: JPG, PNG, MP4, MOV
min_face_sizeintegerNo40Minimum face size in pixels to detect
detection_thresholdfloatNo0.7Confidence threshold for face detection (0.0-1.0)
trackingbooleanNotrueWhether to track faces across video frames

​
Configurations

​
Detection Modes

The extractor supports different detection modes based on use case requirements:

ModeDescriptionBest For
standardBalanced accuracy and performanceGeneral face detection
high-accuracyPrioritizes detection accuracyIdentity verification
high-performancePrioritizes processing speedReal-time applications

​
Configuration Examples

Standard
{
  "detection_mode": "standard",
  "min_face_size": 50,
  "detection_threshold": 0.7,
  "extract_attributes": true
}
High Accuracy
{
  "detection_mode": "high-accuracy",
  "min_face_size": 80,
  "detection_threshold": 0.85,
  "extract_attributes": true,
  "extract_landmarks": true
}
High Performance
{
  "detection_mode": "high-performance",
  "min_face_size": 40,
  "detection_threshold": 0.6,
  "extract_attributes": false
}

​
Attribute Extraction

OptionTypeDefaultDescription
extract_attributesbooleantrueExtract facial attributes (age, gender, emotion)
extract_landmarksbooleanfalseExtract facial landmarks (eyes, nose, mouth positions)
generate_embeddingbooleantrueGenerate face embedding for identity matching
crop_facesbooleantrueGenerate cropped face images

​
Configuration Examples

Sample
{
  "extract_attributes": true,    // Extract age, gender, emotion
  "extract_landmarks": true,     // Extract facial landmarks
  "generate_embedding": true,    // Generate face embedding
  "crop_faces": true             // Create cropped face images
}

​
Processing Flow

​
Output Schema

This feature extractor will output as features in the feature store.

{
  "document_id": "doc_abc123",
  "collection_id": "col_xyz789",
  "source_object_id": "obj_def456",
  
  // Face detection results
  "face": {
    "id": "face_12345",
    "confidence": 0.96,
    "bounding_box": {
      "x": 125,
      "y": 80,
      "width": 120,
      "height": 150
    },
    "attributes": {
      "age_estimate": 32,
      "gender": "female",
      "emotion": "happy",
      "confidence": 0.89
    },
    "embedding": {
      "model": "facenet_v2",
      "dimension": 512,
      "vector": [0.12, 0.34, ...], // Truncated for brevity
      "normalized": true
    },
    "cropped_face_url": "https://storage.example.com/faces/face_12345.jpg"
  },
  
  // Media metadata
  "media": {
    "type": "image", // or "video"
    "filename": "team_photo.jpg",
    "width": 1920,
    "height": 1080,
    "timestamp": null, // For video, this would be the frame timestamp
    "frame_number": null // For video, this would be the frame number
  }
}

Was this page helpful?

Suggest editsRaise issue