Powered by pgvector + GPU acceleration

Face Recognition API Made Simple

Detect, encode, compare, and search faces with enterprise-grade accuracy. Build powerful identity verification and facial analysis features in minutes, not months.

<10ms
Search latency
99.7%
Detection accuracy
1M+
Faces indexed

Features - Everything you need for face recognition

A complete API for face detection, encoding, comparison, and large-scale search. Built for developers who need production-ready face recognition.

Face Detection

Detect multiple faces in images with precise bounding boxes, landmarks, and confidence scores.

Face Encoding

Generate 128-dimensional face embeddings for identity matching using dlib's ResNet model.

Face Comparison

Compare two faces and get similarity scores with configurable thresholds for verification.

FaceSet Search

Search millions of faces in milliseconds using pgvector HNSW indexes for similarity search.

Real-time Tracking

Track faces across video frames with correlation-based tracking and identity persistence.

Enterprise Security

OAuth 2.0 authentication via Keycloak, API keys, rate limiting, and audit logging.

GPU Accelerated

CUDA-powered processing for blazing fast face detection and encoding.

Vector Database

PostgreSQL with pgvector and pgvectorscale for production-grade vector search.

Simple Integration - Up and running in minutes

A clean REST API that works with any language or framework. Get started with a single API call.

# Detect faces in an image
curl -X POST "https://sight.b9cloud.com/api/v1/detect" \
  -H "X-API-Key: your-api-key" \
  -F "image=@photo.jpg"

# Response
{
  "faces": [
    {
      "box": [120, 80, 280, 320],
      "confidence": 0.9987,
      "landmarks": {...}
    }
  ],
  "count": 1
}

Ready to get started?

Sign up for free and start building with BrightSight in minutes. No credit card required.