SightRadar

SightRadar is a face recognition API that lets teams index, search, compare, and detect faces—built for production use at scale. You can find a person across thousands of photos from a single selfie, run 1:1 verification by setting the similarity threshold you want, or detect and quality-gate faces without persisting images.

It is compatible with AWS Rekognition request and response shapes for common operations such as CreateCollection, IndexFaces, SearchFacesByImage, CompareFaces, and DetectFaces. This means you can migrate by pointing your existing AWS Rekognition SDK at a new endpoint URL and keep your existing call patterns (including threshold-based verification via CompareFaces).

SightRadar supports a full workflow: index photos into collections (faces are detected, embedded, and stored), then search a gallery from a selfie to retrieve matches, or compare two faces for similarity scoring. It also handles edge cases: a photo with 0 detected faces is treated as a valid result rather than an error—so you can design your pipeline confidently.

Billing is pay-as-you-go, billed per photo processed (never per face), with no subscription and no minimums. Real-time is $0.00093 per photo processed, and batch indexing is $0.00062 per photo processed. You can also start free with trial credits and no credit card; control-plane calls are free. Storage for search by face-id is $6 / M / mo.

What can I do with SightRadar?

IndexFaces to build collections, SearchFacesByImage to find matches in a gallery, CompareFaces for 1:1 verification using a similarity threshold, and DetectFaces to locate/quality-gate faces without persisting images.

Is SightRadar compatible with AWS Rekognition so I can avoid a rewrite?

Yes. SightRadar uses the same request and response shapes for Rekognition-style operations (e.g., CreateCollection, IndexFaces, SearchFacesByImage, CompareFaces, DetectFaces), so you can swap the endpoint URL and keep using the same SDK calls.

Can I run 1:1 face verification with my own acceptance threshold?

Yes. Use CompareFaces and set the threshold you want to enforce. The response includes similarity scores so you can tune your acceptance logic.

Does SightRadar bill per face or per photo?

Per photo processed. Billing is never per face, and there’s no subscription or minimums.

What happens if an image has no detectable faces?

It’s treated as a valid result. A 0-face photo won’t fail your workflow—you can handle empty matches safely.

How does data retention work for detection and comparison?

Detect and compare process images in memory and don’t persist them. Indexed face data lives in collections you can delete via the API or console.