Find cross-session image similarity matches | Verifa

Compares face embeddings stored for the given identity against all other embeddings in the organization. Returns identities with similar selfies or document photos, ranked by similarity score.

This endpoint detects reused photos and documents across sessions — a key signal for synthetic identity fraud, document reuse, and fraud rings.

Results are deduplicated by identity (keeping the highest similarity score when multiple embeddings match). Only matches above the specified threshold are returned.

Compares face embeddings stored for the given identity against all other embeddings in the organization. Returns identities with similar selfies or document photos, ranked by similarity score. This endpoint detects reused photos and documents across sessions — a key signal for synthetic identity fraud, document reuse, and fraud rings. Results are deduplicated by identity (keeping the highest similarity score when multiple embeddings match). Only matches above the specified threshold are returned.

Authentication

X-API-Keystring

Organization API key. Keys are prefixed with vk_live_ (production) or vk_sandbox_ (sandbox).

Path parameters

identity_idstringRequired

Identity ID (idn_*) or external reference.

Query parameters

thresholddoubleOptional

Minimum cosine similarity score (0.1–1.0) to count as a match. Default 0.6 balances precision vs recall. Use 0.8 for high-confidence matches only; use 0.4 for broader search with more false positives.

source_typeenumOptional

Filter by embedding source type. selfie matches selfie-to-selfie; document_front matches ID photo-to-ID photo. If omitted, all source types are compared.

Allowed values:

limitintegerOptional1-100Defaults to 20

Maximum number of matches to return.

offsetintegerOptional>=0Defaults to 0

Offset for pagination.

Response

Paginated list of image similarity matches.

datalist of objects

paginationobject

Errors

401

Unauthorized Error

404

Not Found Error

1	{
2	"data": [
3	{
4	"identity_id": "idn_def456",
5	"session_id": "ses_xyz789",
6	"similarity_score": 0.8234,
7	"matched_at": "2026-03-01T12:00:00Z",
8	"embedding_id": "facemb_abc123"
9	}
10	],
11	"pagination": {
12	"total": 1,
13	"limit": 1,
14	"offset": 1
15	}
16	}