Skip to main content

Under-eye check

Dark Circle Detector From a Photo

Updated 3 min read

The Dark circles analysis finds both eyes with facial landmarks, samples the skin just below them and compares how much darker it is, then reports a score and a band from Minimal to Severe with a highlighted overlay. It is a quick, private way to compare photos over time, for example before and after a few nights of better sleep or a change of skincare.

It is a cosmetic estimate from pixel brightness, not a medical assessment. Everything runs in your browser through face-api.js on TensorFlow.js. The photo is never uploaded, and recent scans keep only the numbers, never the image.

Want every option? Open the Dark circles analysis

How to use it

  1. 1

    Take a front-facing photo without makeup under the eyes, in soft daylight from a window in front of you.

  2. 2

    Add it to the Dark circles analysis and wait for both eye regions to be located with landmarks.

  3. 3

    Read the severity score and band, and check the overlay to make sure it sampled the area under each eye.

  4. 4

    Repeat in the same spot and light on another day so the 2 photos can be compared fairly.

How the dark circle check works

After finding the face, the tool places 68 landmark points and uses the six around each eye to locate the eye openings. It then samples a region just below each eye and measures its brightness pixel by pixel, skipping pixels that are almost pure black or white because those are usually shadows or reflections. The darker that region reads, the higher the score.

The score is grouped into five bands: Minimal, Mild, Moderate, Significant and Severe. The overlay shows which pixels were sampled, so you can check that the tool looked at your under-eye area rather than a lash line or the frame of your glasses.

Lighting decides most of the result

Because the check works from brightness, the light in the photo matters more than anything else. Overhead light casts the brow's shadow straight into the under-eye area and makes almost anyone look tired. Side light darkens one eye more than the other. Soft daylight from in front of you, with no strong lamp above, gives the most honest reading.

Skin tone matters too. The measure is brightness, and natural pigmentation around the eyes differs between people, so the score is most useful for comparing your own photos under the same conditions, not for comparing different people.

  • Face a window; avoid overhead ceiling lights.
  • Remove concealer and tinted eye cream for a fair comparison.
  • Keep the camera at eye level so the brow does not shade the eyes.
  • Glasses frames and lenses can cast shadows or reflections in the sampled area.
  • Use the same place, time of day and distance when comparing photos.

What it cannot tell you

Under-eye darkness has many causes: genetics, thin skin, pigmentation, hollows that cast shadows, allergies, rubbing the eyes, dehydration and sleep. The tool sees only that a patch of pixels is darker than expected; it cannot tell which cause applies, and a high score is not a diagnosis. If darkness or swelling around the eyes appears suddenly, is on one side only or comes with other symptoms, talk to a doctor or pharmacist rather than relying on a photo tool.

Frequently asked questions

How does the dark circle detector measure darkness?

It locates each eye with facial landmarks, samples the skin just below it and measures how dark those pixels are, then turns that into a score and a severity band.

Why did I get a high score after a good night's sleep?

Overhead or side lighting, glasses shadows and natural pigmentation can all raise the score. Retake the photo facing soft daylight with the camera at eye level.

Can I compare results over time?

Yes, if you keep the light, place, distance and camera the same. Different conditions change the score more than most real changes do.

Is this a medical test?

No. It is a cosmetic estimate from pixel brightness. It cannot identify causes or conditions. See a professional for anything that worries you.

Does it work for all skin tones?

It measures brightness, and natural under-eye pigmentation varies between people, so scores are not comparable between individuals. Compare your own photos instead.

Is the photo uploaded?

No. The analysis runs entirely in your browser.

Run a website? Link to this guide if it helps your readers, and see our free backlink sites list for places to get your own site listed.