Shape Detection: FaceDetector and TextDetector
Finds faces and printed text in an image (experimental flag only).
In one line
Can find faces and printed text in a picture, but only with a special setting on
Finds faces and printed text in an image (experimental flag only).
Face boxes and text blocks from images without server calls. Experimental and off by default; not a stable surface.
What it can and can't do
What the user does: You turn on a flag in Chrome, then pick or load a picture.
It can
- Find where faces are in a picture, without naming them
- Find where printed text sits in a picture
- Work on a photo or a camera frame the page already has
It cannot
- Work in normal Chrome, Edge, Firefox or Safari without the flag turned on
- Tell who a person is
- Work on Android or iPhone browsers
- Work on pictures the page does not have
How it differs between platforms: Only desktop Chrome and Edge with a flag turned on. No phone browsers.
Browser support
| Browser | Status | First version | Note |
|---|---|---|---|
| Chrome | Behind a flag | Not available | Experimental flag only |
| Edge | Behind a flag | Not available | Experimental flag only |
| Firefox | Not supported | Not available | None |
| Safari | Not supported | Not available | None |
| Safari on iPhone | Not supported | Not available | None |
| Chrome on Android | Not supported | Not available | None |
| Samsung Internet | Not supported | Not available | None |
How a site asks for it
| API | new FaceDetector().detect(img), new TextDetector().detect(img) |
| How access is granted | settings |
| What the user sees | Flag chrome://flags/#enable-experimental-web-platform-features |
| Needs a secure context (HTTPS) | Yes |
| Needs a user gesture | No |
| How long it lasts | Per origin |
| How to take it back | Site settings |
| Inside an iframe | Same as top level; cross-origin iframes need a permissions policy where listed |
Typical flow
- Call the API
- Browser applies its access rules
- Page receives the result
Data you get
What comes back: Finds faces and printed text in an image (experimental flag only).
Shape: object
How often: on demand
| Field | Type | Unit | Range | Precision |
|---|---|---|---|---|
boundingBox | object | pixels | image rect | Not listed |
landmarks[] | list | pixels | eyes, nose, mouth | Not listed |
text (TextDetector) | string | None | recognised text | Not listed |
Technical limits
- No stable release in any browser
- No BCD entry
The story
What it is
A page can find faces and printed words in a picture, but only in a test mode.
For you
You will not meet it on normal sites. It is behind a flag, so only people who turn it on see it.
For makers
Makers would use it for photo tags, camera filters and reading text from pictures. It is not on by default because face finding raises privacy worries.
Where you see it
| Type | Example | Apps |
|---|---|---|
| creative | A photo editor crops a picture around the faces it finds. | None named |
| social | A camera filter app places stickers on faces. | None named |
| education | A tool reads text on a worksheet photo for a translator. | Google Translate |
| accessibility | A page reads out the text in a picture to a blind user. | None named |
| security | A bad site could count faces in a photo you upload. | None named |
| games | A browser game moves a character as you turn your head. | None named |
Risks
- Face scanning
- Not widely supported
What the platform says
- Google: Shape Detection API. Chrome's docs describe detecting faces, barcodes and text in images, with faces and text still behind a flag.
Good uses and abuse
Legitimate uses
- Photo tagging
- Accessibility
Known abuse
- Face detection without consent
Scores
| Score | Value | Meaning |
|---|---|---|
| Sensitivity | 3 of 5 | How revealing the data is |
| Utility | 2 of 5 | How useful it is to honest apps |
| Friction | 3 of 5 | How much the user has to do |
| Fingerprint bits | Not rated | Rough identifying entropy, when known |
History
- 2018, Chrome: Introduced behind experimental flag; no stable release since
Sources
How well checked: Inferred from related behavior.