Machine Learning Engineer and Founder
Ahtisham ul Haq builds the computer vision pipeline behind The Face Shape Detector. He works in machine learning and deep learning, with convolutional neural networks and facial landmark regression as his day-to-day subject, and he wrote the measurement methodology that every tool on this site reports against.
Experience and credentials
Machine learning and deep learning engineer specializing in convolutional neural networks and computer vision. Author of the site measurement methodology and of the landmark pipeline that reads 468 facial points in the browser.
Writes about
6
Published articles
65
Minutes of reading
3
Categories written in
5
Areas of focus
Published work
Newest first. Every one is measured against a number rather than a trend.
A man's apparent face shape is built by the beard line and the hairline as much as by bone. This guide covers cuts, beard styles, frames and maintenance lines for all seven shapes.
Well-known faces make the seven shape categories legible in a way that diagrams cannot. Use these commonly cited examples to calibrate your eye, then measure your own outline instead of matching a lookalike.
A complete hair guide organized by face shape, from the geometry that governs every cut to named styles for all seven shapes. Includes fringe matching, texture adjustments and exactly what to tell your stylist.
Frames work by contrast, and frame width should match the widest part of your face. This guide covers the right styles for all seven face shapes, plus color and fit.
A face shape detector is a short pipeline of well-understood vision components, not magic. Here is how 468 landmark points become normalized ratios and a confidence score.
Four readings and two ratios are all it takes to name your face shape without guesswork. This guide covers the protocol, the thresholds, and a worked example.
Keep exploring
The reference pages and tools these articles point back to.
The advice in these articles is written for a shape. Find out which one is yours first.