About Face Detector and Face Blur
This face detector is free, needs no sign-up and runs in your browser, so your photos and camera frames are never uploaded. Add a picture, or start your webcam, and it draws a box and six keypoints (eyes, nose tip, mouth and ears) on every face it finds, with a confidence score. The count of faces is shown above the picture.
Switch What to draw to Blur faces or Pixelate faces and every detected face is hidden. The region is grown by 25 percent first so the hairline, chin and ears are covered too. The hiding is baked into the PNG you download, so the face pixels are not in the file, not just painted over a layer.
It uses the MediaPipe Face Detector with the BlazeFace short range model, from Google, released under Apache-2.0. The model file is tiny, about 230 KB, but MediaPipe also needs its WebAssembly runtime of about 13 MB, so the card shows about 13 MB in total. The runtime is served by tool.ulkaa.com itself and only the model file comes from Google model storage, once, and your browser may cache both. Your pictures and video never leave your device.
The short range model is made for faces within about two metres of the camera, such as selfies, group photos and webcams. Small faces in a wide crowd shot, faces turned fully away, and heavy occlusion may be missed. Always look at the result before you rely on a blur for privacy.
How to use Face Detector and Face Blur
- Choose Image or Webcam as the source.
- The small model starts loading in the background when the page opens, unless Data Saver is on, and the card shows its progress and tells you when you can start.
- Add a picture by dropping it, tapping the box, pasting it or using the phone camera, or press Start camera.
- Choose Show boxes to see detections, or Blur faces or Pixelate faces to hide them. Use the strength slider to make the hiding stronger.
- Lower Minimum confidence if a face is missed, or raise it if something that is not a face gets a box.
- Press Download PNG to save the picture exactly as it appears, at its full size.
Features
- Face boxes with six keypoints and a confidence score for each face.
- A live count of the faces found.
- Blur or pixelate every face, with a strength slider, baked into the downloaded PNG.
- Detection regions are grown by 25 percent so hair and ears are covered too.
- Image input by drop, file choice, paste or phone camera, plus a live webcam mode.
- Webcam frames are drawn in an animation loop, without re-rendering the page for every frame.
- An adjustable minimum confidence.
- Photos larger than 4096 px on the long side are scaled down to that for detection and download.
How a face region is hidden
The detector returns a box for each face. The tool grows it by 25 percent on every side, clamps it to the picture, then either shrinks the region and stretches it back smoothed (blur) or replaces each square block with its average colour (pixelate). The block size is about a tenth of the region times the strength.
region = box grown by 25 percent, clamped to the picture
block = max(4, round(short side of region / 10 x strength))Worked examples
A face box at x 100, y 80, 120 wide and 150 tall in a 640 by 480 picture gives Hidden region: x 70, y 42, 180 by 226
At strength 1.5 the block size is 27 px.
A face box near the left edge: x 5, y 10, 100 by 100 gives Hidden region: x 0, y 0, 130 by 135
The region is clamped to the picture, so it never reaches outside it.
A group photo with three faces gives Faces found: 3
The counter shows how many faces passed the minimum confidence.
Good to know
Detection is not recognition
This tool finds where faces are. It does not work out who anyone is, and it does not store or compare faces. The result is a set of boxes and keypoints that exist only in your tab.
Blurring and privacy
A blur is only as complete as the detection. A face the model misses stays visible, so check the whole picture before you share it, and lower the minimum confidence if a face is not boxed. Pixelate with a large block size is the more robust way to hide a face, because a mild blur can sometimes be partly reversed.
The downloaded PNG has no metadata copied from your original, because it is drawn fresh on a canvas.
Privacy: how it works
- Pictures and camera frames are processed in your browser and are never uploaded. The camera stream is not recorded or stored and stops when you press Stop camera or leave the page.
- Two downloads are needed: the BlazeFace model file (about 230 KB) from Google model storage, and the MediaPipe runtime (about 13 MB) from tool.ulkaa.com itself. Neither request contains your pictures.
Tips
- Use Pixelate for faces that must not be recognised, and a stronger setting than you think you need.
- For a crowd shot, lower the confidence to 30 percent and check each box.
- Hold the phone camera steady and in good light for the best live detection.
- Use the counter to check you caught everyone before you blur.
Frequently asked questions
Is it really free?
Yes. There is no price, no trial and no credit card. Ulkaa Tool does the work with your own device, so there is no server bill to pass on to you.
Do I need an account?
No. There is no sign-up and no login. Open the page and use it. An Ulkaa account is not needed for anything on this tool.
Does it run in my browser?
Yes. Your pictures and camera frames are processed by code running in your browser tab, on your own device.
Is my data uploaded?
No. Your pictures and camera frames are never sent to Ulkaa or to anyone else. You can check this yourself by watching the Network tab in your browser developer tools while you use the tool.
How do I blur faces in a photo?
Add the photo, set What to draw to Blur faces or Pixelate faces, check that every face is covered, then press Download PNG.
Is the blur permanent?
Yes. The hiding is drawn into the pixels of the PNG you download, so the original face pixels are not in the file.
Which model does it use?
MediaPipe Face Detector with the BlazeFace short range model, from Google, under Apache-2.0.
What is downloaded, and from where?
The tiny model file comes from Google model storage, and the 13 MB MediaPipe runtime is served by tool.ulkaa.com itself. Your pictures are never part of those requests.
Can it identify who is in the photo?
No. It only finds faces and does not recognise or compare them.
Why was a face missed?
The short range model works best for faces within about two metres. Very small, side-on, covered or blurred faces can be missed. Lower the minimum confidence and check again.
Does the webcam video get recorded?
No. Frames are drawn to a canvas and thrown away. The camera stops when you press Stop camera.
Popular Face Detector and Face Blur pages
- Blur Faces in a PhotoHide every face in a picture before you post or send it.
- Pixelate Faces in a PhotoReplace each face with coarse blocks so it cannot be recognised.
- Face CounterCount the faces in a picture and see each one outlined.
- Webcam Face DetectionSee faces found live in your camera, with boxes and keypoints.