API Reference — Face Recognition

Manajemen Pipeline

Endpoint untuk membuat dan menghapus pipeline Face Recognition pada stream video. Pipeline NFV4-FR menjalankan deteksi dan recognition standar; NFV4-FRA menambahkan analisis atribut demografis (gender, estimasi usia, pakaian) secara paralel.

curl -X POST "http://localhost:4004/pipeline/0/5136c6c0916cd30c/NFV4-FR" \  -H "Content-Type: application/json" \  -d '{    "fr_primary_threshold": 0.8,    "fr_secondary_threshold": 0.4,    "mode": 0,    "similarity_threshold": 0.85,    "fr_min_face_size": 48,    "frontal_only": true,    "sampling_time_ms": 50,    "fr_timeout_dump": 1500,    "fr_sampling_size": 8,    "static_object_sensitivity": 5,    "fr_blur_filter_threshold": 4  }'

{  "analytic_id": "NFV4-FR",  "code": 200,  "message": "pipeline successfully created",  "node_num": 0,  "stream_id": "5136c6c0916cd30c"}

{  "code": 400,  "message": "Invalid request"}

{  "code": 409,  "message": "Pipeline already exists"}

{  "code": 500,  "message": "Internal server error"}

curl -X DELETE "http://localhost:4004/pipeline/0/5136c6c0916cd30c/NFV4-FR"

{  "analytic_id": "NFV4-FR",  "code": 200,  "message": "pipeline successfully deleted",  "node_num": 0,  "stream_id": "5136c6c0916cd30c"}

{  "code": 404,  "message": "Stream not found"}

curl -X POST "http://localhost:4004/pipeline/0/5136c6c0916cd30c/NFV4-FRA" \  -H "Content-Type: application/json" \  -d '{    "mode": 1,    "similarity_threshold": 0.75,    "fr_primary_threshold": 0.7,    "frontal_only": false,    "fr_min_face_size": 48,    "fr_max_fps": 5,    "fr_sampling_size": 5  }'

{  "analytic_id": "NFV4-FRA",  "code": 200,  "message": "pipeline successfully created",  "node_num": 0,  "stream_id": "5136c6c0916cd30c"}

{  "code": 400,  "message": "Invalid request: missing required field 'keyspace'"}

{  "code": 400,  "message": "Invalid request: missing required field 'keyspace'"}

curl -X DELETE "http://localhost:4004/pipeline/0/5136c6c0916cd30c/NFV4-FRA"

{  "analytic_id": "NFV4-FRA",  "code": 200,  "message": "pipeline successfully deleted"}

{  "code": 400,  "message": "Invalid request: missing required field 'keyspace'"}

Enrollment & Pencarian Wajah

Operasi enrollment, recognition, dan pencarian embedding kini didokumentasikan secara penuh di section FREMIS-N. Visionaire4 meneruskan request ke service FREMIS-N — gunakan basis URL FREMIS-N (FREMIS_N_API_URL) untuk memanggil endpoint berikut secara langsung:

Event Streaming (WebSocket)

Visionaire4 mengirim setiap event hasil deteksi/inferensi melalui WebSocket. Untuk berlangganan event yang dihasilkan oleh pipeline Face Recognition pada stream tertentu, buka koneksi WebSocket ke endpoint berikut:

Pipeline NFV4-FR (face recognition standar):

ws://<host>:4004/event_channel/{node_num}/{stream_id}/NFV4-FR

Pipeline NFV4-FRA (face recognition + atribut demografis):

ws://<host>:4004/event_channel/{node_num}/{stream_id}/NFV4-FRA

Gunakan NFV4-FR jika Anda hanya membutuhkan event identitas wajah. Gunakan NFV4-FRA jika pipeline juga menjalankan analisis atribut (gender, estimasi usia, pakaian) dan Anda ingin menerima field atribut dalam pipeline_data.

Setelah koneksi terbuka, server akan langsung mendorong (push) setiap event ke client tanpa perlu handshake atau pesan subscribe. Tutup koneksi untuk berhenti menerima event.

Parameter Path

Query Parameter (Opsional)

Format Pesan

Setiap event yang diterima berbentuk JSON dengan struktur berikut:

{
  "event_id": "evt_123456",
  "stream_id": "<stream_id>",
  "analytic_id": "NFV4-FR",
  "node_num": 0,
  "timestamp_ms": 1674123456789,
  "primary_text": "John Doe",
  "secondary_text": "Employee",
  "pipeline_data": {
    /* field spesifik module — lihat "Output yang Dihasilkan" pada index.mdx */
  },
  "image_jpeg": "base64-encoded JPEG..."
}

Detail field pipeline_data untuk Face Recognition dijelaskan pada Output yang Dihasilkan di halaman utama.

Contoh Koneksi

const ws = new WebSocket(
  'ws://localhost:4004/event_channel/0/<stream_id>/NFV4-FR'
);

ws.onmessage = (msg) => {
  const event = JSON.parse(msg.data);
  console.log(event.primary_text, event.pipeline_data);
};

ws.onerror = (err) => console.error('WebSocket error', err);
ws.onclose = () => console.log('Disconnected');

import asyncio
import json
import websockets

async def listen():
    url = "ws://localhost:4004/event_channel/0/<stream_id>/NFV4-FR"
    async with websockets.connect(url) as ws:
        async for msg in ws:
            event = json.loads(msg)
            print(event["primary_text"], event["pipeline_data"])

asyncio.run(listen())

Filter Berdasarkan Logic

Jika pipeline menjalankan beberapa logic rule, Anda dapat memfilter event hanya untuk satu logic dengan menambahkan query parameter:

ws://<host>:4004/event_channel/{node_num}/{stream_id}/NFV4-FR?logic=<logic_name>

Endpoint Channel Lainnya

Untuk mendengarkan event yang lebih luas (lintas analitik atau lintas stream), gunakan endpoint berikut yang dijelaskan pada Developer Guide: