Eye Shape Detector
Find your eye shape in three ways: live camera, a photo upload, or a guided measurements quiz. No signup, and your photo is never uploaded or stored.
Eye Shape Detector Tool
Best results: one front-facing face, eyes open and relaxed, looking at the camera, even light, no sunglasses or hair over the eyes.
1 Start the camera · 2 Line your face up with the oval until the checks turn green · 3 Capture. The frame is analysed on your device and never uploaded.
Start the camera to get live positioning tips.
Camera not working? instead.
Drag & drop a front-facing photo here, or click to browse
JPG, JPEG, PNG, or WEBP, under 10 MB. One clear, front-facing face works best.
This quiz is not AI image analysis. It classifies from your own observations, so it suits you if you do not want to use a camera or photo, and it answers the crease and depth questions a 2-D photo cannot answer reliably.
Your result
—
Also present / overlapping traits
Confidence: —
Measurements behind this result
How this was determined
The card shows your result and is branded OvalFaceShape.com Eye Detector · Built by Rehan Mir. Your photo is not included.
No beauty ranking or attractiveness score is given. If the photo was angled, shadowed, or the eyes were partly closed, treat the result as approximate and check it with the quiz. Curious about the eye type behind the look? Read our hunter eyes guide.
What is an eye shape detector?
An eye shape detector looks at the outline of the eye opening, the eyelid crease, the position of the inner and outer corners, and how much of the iris is covered by the lids. It then matches those visible traits to a named category such as almond or round. It is not a medical test, an identity check, or a beauty score.
How do I tell what shape my eyes are? Look straight ahead with a relaxed face. Check the crease, the white around the iris, and whether the outer corner is higher, lower, or level with the inner corner. The tool above does the same checks from measurements, and the quiz asks you to make them yourself.
How the Eye Shape Detector works
Landmarks first, label second
For camera and photo methods, the browser loads MediaPipe Face Landmarker (478 points: 468 face points + 10 iris points) and reads only eye geometry. No separate “AI score” is invented.
Measurements
Per eye: width = inner to outer corner; height = upper to lower lid; tilt = outer corner height relative to the inner corner; iris exposure = white above or below the iris; spacing = inner-corner gap ÷ eye width.
Transparent rules
Rules and thresholds are listed in the Accuracy section. A result is labelled Strong, Likely, or Approximate from how clearly the measurements clear a threshold and how well the two eyes agree — never 100%.
What a photo cannot do
A single 2-D photo cannot reliably judge a hidden crease, true depth, or brow-to-lid fold. Those traits are not claimed from a photo; the quiz asks you directly.
One face at a time
The model is set to one face. If more than one face is in frame, move closer, crop to one face, and re-analyse. If no face is found, you get an error, not a guessed shape.
Shown, not hidden
The result draws the corner line and iris circle it used on your photo and lists every number, so you can see exactly what drove the label.
Three ways to identify your eye shape
1. Live camera
Best for a quick check when you are alone, in good light, facing the camera. Permission is asked only after you tap Start camera, and you can stop the camera at any time.
Use camera2. Upload a photo
Best for a photo you already trust. JPG, JPEG, PNG, or WEBP under 10 MB. The page draws the eye landmarks it used, so you can see what was measured.
Upload photo3. Measurements & quiz
Best if you prefer no photo, or if crease and depth are the deciding traits. It is clearly labelled as a quiz result, not AI image analysis.
Take quizDifferent types of eye shapes
Categories overlap. A person can be almond-shaped and hooded and slightly upturned. Read the primary shape together with tilt, spacing, and depth — not as one exclusive box.
Almond Eyes
Defining traits: wider than tall, tapered at both corners, visible crease, iris touches upper and lower lids.
Common mistake: calling any long eye almond when white is clearly visible above or below the iris — that openness points toward round.
Round Eyes
Defining traits: height closer to width, rounded outline, white visible above or below the iris when looking straight ahead.
Common mistake: judging roundness from a photo taken from below or with raised brows, which opens the eye artificially.
Monolid Eyes
Defining traits: no visible crease; the lid looks smooth from lash line to brow.
Common mistake: confusing monolid with hooded. Monolid has no crease to hide; hooded has a crease hidden by a fold.
Hooded Eyes
Defining traits: a fold of skin partly or fully covers the crease when eyes are open, so visible lid space is small.
Common mistake: deciding hooded from a photo with eyes looking down, which also hides the lid.
Upturned Eyes
Defining traits: outer corner sits higher than the inner corner — a positive corner tilt. Strongly upturned, slightly hooded eyes are often called hunter eyes.
Common mistake: tilting the head or camera, which moves both corners together and creates a false tilt.
Downturned Eyes
Defining traits: outer corner sits lower than the inner corner — a negative corner tilt.
Common mistake: reading downturn from tired or half-closed eyes instead of the relaxed, open position.
Deep-Set Eyes
Defining traits: eyes sit back in the socket under a prominent brow bone, often with shadow above the lid. A depth trait, best judged in person or in the quiz, not from one flat photo.
Protruding Eyes
Defining traits: eyes sit forward in the socket, with a larger visible eye surface and a fuller lid curve. Depth cannot be proved from a single 2-D photo, so the quiz is the deciding method.
Find your shape in 3 checks
Face a mirror, head level, eyes open and relaxed. Answer in order.
Eyelid crease
Iris and white
Corner tilt
What the detector measures
- Width — inner to outer corner. The base for every ratio.
- Height — tallest visible lid opening. Height ÷ width separates almond from round.
- Tilt — angle of the corner line against level. Positive is upturned, negative is downturned.
Eye spacing ratio
Eye spacing ratio compares the gap between your inner eye corners with the width of one eye. It decides whether your eyes read as close-set, average, or wide-set, and it applies on top of every base shape above.
How the ratio is measured
Spacing ratio = distance between the two inner corners ÷ average eye width. The “five-eye” layout shown is a common teaching guideline for an average ratio near 1.0. It is not a medical standard or a beauty score.
Read the Eye Spacing Ratio guideClose-set
The gap between the eyes is clearly narrower than one eye width.
Average spacing
The gap is about one eye width.
Wide-set
The gap is clearly wider than one eye width.
Thresholds are this tool’s own heuristics, and the detector above reports your spacing ratio from a photo.
Eye shape comparison chart
| Shape | Key check | Crease | Corners | Iris |
|---|---|---|---|---|
| Almond | Wider than tall, tapered corners | Visible | Level or slightly up | Touches both lids |
| Round | Height close to width, open outline | Visible | Level | White above / below |
| Monolid | Smooth lid surface | None visible | Any | Varies |
| Hooded | Fold covers crease, little lid visible | Hidden | Any | Varies |
| Upturned | Outer corner higher | Varies | Up | Varies |
| Downturned | Outer corner lower | Varies | Down | Varies |
| Deep-set | Recessed under brow | Varies | Any | Often partly shadowed |
| Protruding | Forward, fuller surface | Varies | Any | More fully visible |
A teaching summary of the definitions used by this tool, not a clinical diagnostic table. Thresholds are the tool’s own heuristics (see Accuracy).
How to identify your eye shape manually
- Face a mirror, head level, face relaxed, eyes open, looking straight ahead. Do not raise your brows or squint.
- Crease first: none = monolid; hidden by a fold = hooded; clearly visible = almond, round, or another open shape.
- Iris next: touches both lids = almond-like; white above or below = round-like.
- Corners next: draw an imaginary line from inner to outer corner. Outer above = upturned; below = downturned; on the line = level.
- Spacing and depth last: compare the inner-corner gap with one eye width, and check depth from the side. These are traits, not a replacement for the base shape.
Common identification mistakes
- Judging from an angled selfie, a photo from below, or with brows raised.
- Confusing eye colour, eyebrow shape, or face shape with eye shape.
- Forcing one label when two traits overlap — for example, almond + hooded + upturned.
- Deciding hooded or deep-set from a single flat photo, where depth and folds are hard to see.
- Using a photo with eyes half-closed, in shadow, or behind glasses glare.
Eye shape vs. eye colour vs. eyebrow shape
Rarest eye shape, eyebrow, eyelashes and face shape
Once you know your eye shape, see how rare your features are across the face.
Rarest Eye Shape
Find out which eye shape is considered the rarest and how it compares with the common ones.
Read the guide →Rarest Eyebrow Shape
See which eyebrow shape is rarest and how brow shape changes the look of your eyes.
Read the guide →Rarest Eyelashes Types
Explore the rarest eyelash types and what sets them apart.
Read the guide →Rarest Face Shape
Learn which face shape is rarest and how it differs from the common shapes.
Read the guide →Accuracy and limitations — the actual rules used
These are this tool’s classification heuristics, not universal scientific thresholds. Eye-shape categories are widely used stylistic categories, not a peer-reviewed medical taxonomy, and different guides draw the lines differently.
- Height ÷ width: ≥ 0.40, or ≥ 0.37 with clear white above/below the iris, leans Round; 0.26–0.38 with the iris touching the lids leans Almond. Between those bands the result is labelled Approximate and the borderline trait is shown.
- Corner tilt: ≥ +3.5° = upturned; ≤ −3.5° = downturned; between = level. Tilt is measured after averaging both eyes; a large left/right disagreement lowers confidence.
- Spacing: inner-corner gap ÷ eye width < 0.85 = close-set; > 1.15 = wide-set; otherwise average spacing.
- Iris exposure: white fraction = visible white above + below the iris, divided by iris diameter. > 0.12 supports a round-like reading.
- Not measured from a photo: crease visibility (hooded/monolid) and true depth (deep-set/protruding). The photo result says so explicitly and points to the quiz.
- Confidence: starts from agreement between the eyes and distance from thresholds, then is reduced for a turned head (unequal eye widths) or a borderline ratio. It is capped below 100% and is an uncertainty label, not a probability from a clinical study.
- Known failure modes: angled head, heavy shadow, glasses glare, closed or squinting eyes, low resolution, and multiple faces. In those cases the tool should warn or refuse, not guess.
Privacy and image processing
- Camera and photo analysis run locally in your browser using a downloaded MediaPipe model. No photo is sent to a server by this page.
- No photo is stored by this page, no account is required, and no API key is exposed (none is used).
- Closing the tab, choosing another image, or tapping Retake clears the previous preview from the tool’s working canvas.
- This page adds no analytics script. Third-party loads are limited to the MediaPipe model and runtime (when you start the camera or analyse an image) and the page fonts (Google Fonts), all disclosed here.
- If the site adds analytics later, it should be privacy-conscious and must not collect photos.
- Camera permission can be revoked in your browser’s site settings at any time. Full details are in our Privacy Policy.
Frequently asked questions
How do I tell what shape my eyes are?
Check crease, iris exposure, and corner tilt in that order, with a relaxed, straight-ahead look. The How it works section lists the exact measurements.
How can I identify my eye shape?
Use the camera, a photo, or the quiz on this page. The quiz is best if you do not want to use a photo, or if crease and depth are the deciding traits.
What are the different types of eye shapes?
Almond, round, monolid, hooded, upturned, downturned, deep-set, and protruding — plus spacing traits (close-set / wide-set). Most people have a combination.
How to identify almond eyes?
Wider than tall, tapered corners, visible crease, iris touching both lids.
How to identify hooded eyes?
A fold covers the crease when eyes are open and little lid shows. Use the quiz for this trait, because a flat photo cannot prove it reliably.
Can I detect my eye shape with a camera?
Yes, on HTTPS with camera permission. This tool analyses the captured frame in your browser and does not upload it.
What is the difference between round and almond eyes?
Round eyes are more open and circular with white often visible above or below the iris. Almond eyes are wider than tall, taper at the corners, and the iris usually touches both lids.
More detectors in our library
The Eye Shape Detector is one of several free face-analysis tools on OvalFaceShape.com. Try the others to see your full face profile.
Face Shape Detector
Upload a photo or use your camera to find your face shape.
Open tool →Face Shape Quiz
Answer a few questions to work out your face shape without a photo.
Open tool →Face Shape Calculator
Enter your measurements and calculate your face shape.
Open tool →Golden Ratio Face Detector
Compare your facial proportions with the golden ratio.
Open tool →Face Symmetry Test
Check how symmetrical the two halves of your face are.
Open tool →Lip Shape Analyzer
Find your lip shape and see the measurements behind it.
Open tool →Explore face shapes and more guides
Once you know your eye shape, read how each face shape works, try the lip shape analyzer, or learn about hunter eyes.
Face shape guides
Oval Face Shape
Read →Oblong Face Shape
Read →Diamond Face Shape
Read →Round Face Shape
Read →Rectangle Face Shape
Read →Square Face Shape
Read →Triangle Face Shape
Read →Inverted Triangle Face Shape
Read →Pear Face Shape
Read →Heart Face Shape
Read →Founder / About the developer
Muhammad Rehan Iqbal — Rehan Mir
Founder & Web Developer, OvalFaceShape.com
Rehan Mir builds the site’s face-analysis tools, including this Eye Shape Detector, and is responsible for its design, measurement rules, and privacy approach described on this page. No medical, scientific, or clinical qualification is claimed here, and this tool is not medical advice. Learn more on the About Us page or get in touch.
External references and sources
- Google MediaPipe — Face Landmarker documentation: the on-device model used for landmark detection (478 points including iris points).
- MediaPipe Face Landmarker model file (float16) loaded by this page from Google Cloud Storage.
- Facial anthropometry landmarks (PMC): terms such as exocanthion (outer corner), endocanthion (inner corner), and eye fissure length used in the manual-measurement guide.
- IPSY — What Is My Eye Shape?: a general beauty-guide summary of the same crease, iris, and corner checks used in the quiz.