What a face detector does
A face detector finds every face in an image and draws a box around each one, with a confidence score. It does not identify anyone. Detection is the first step in any face workflow: you count and locate faces before you compare or search them.
How to read the result
- Faces found is every face the detector located.
- Usable for matching counts faces that also pass the quality gate: large and clear enough to compare reliably. Green boxes pass, red boxes do not.
- Confidence is how sure the detector is that a box holds a face.
If a face you expected is missing or red, it is usually too small, blurred or turned away. The thresholds guide covers how quality and similarity scores feed a match decision.
Common uses
- Reject uploads with no face, or more than one, before an identity check.
- Check that a profile photo or selfie is clear enough to match later.
- Count people in event photos before indexing a gallery for face search.
- Crop or blur faces for privacy using the returned boxes.
Detect faces with the API
This page calls POST /v1/detect, the SightRadar equivalent of AWS Rekognition DetectFaces. It costs 93 credits per photo whatever the number of faces, and stores nothing. See the detect endpoint reference, or compare two faces with the companion tool.
curl -X POST "https://api.sightradar.com/v1/detect" \ -H "Authorization: Bearer $SR_API_KEY" \ -F "file=@photo.jpg"
Frequently asked questions
- What is face detection?
- Face detection finds where faces are in an image. It returns a box around each face and a confidence score, without saying who anyone is. It is the first step before face comparison or face search.
- What is the difference between faces found and faces usable for matching?
- Faces found counts everything the detector located. Faces usable for matching counts the faces that also pass the quality gate, meaning they are large and clear enough to compare reliably. Tiny, blurred or heavily turned faces are found but not counted as usable.
- Are my photos stored?
- No. Detection runs in memory and the image is discarded once the result is returned. No photo, face crop or faceprint is saved, and nothing is used to train models.
- Which files can I upload?
- JPG, PNG or WebP images up to 10 MB. The file type is checked from the image data itself, not just the file name.
- Is the face detector free?
- Yes. Sign in with a free account and each photo is charged against the free credits on that account (93 credits per photo, whatever the number of faces). There is a fair-use limit per visitor.
- Can I detect faces from my own app?
- Yes. The same operation is POST /v1/detect in the SightRadar API, the equivalent of AWS Rekognition DetectFaces. It accepts an upload or an image URL and stores nothing.