CPTPro

Image processing tool that extracts facial features to generate personalized color palettes using color harmony theory and KMeans clustering.

An AI color analysis system that uses Dlib landmark detection and KMeans clustering to extract skin tones, eye color, and hair color from uploaded photos. A color harmony engine maps these values to a personalized palette using seasonal color theory.

How it works

The pipeline takes an uploaded image, runs face detection, then extracts dominant colors from three regions — skin, eyes, and hair — using KMeans clustering on the RGB values. Those clusters are then matched against a seasonal color theory model (Spring, Summer, Autumn, Winter) to produce a curated palette recommendation.

Technical details

• Face detection: Dlib 68-point landmark model localizes facial regions precisely
• Color extraction: KMeans (k=5) on each region’s pixels; picks the most saturated non-background cluster
• Harmony mapping: Lab color space distance used to match extracted colors to seasonal palettes
• Serving: Flask REST API accepts base64-encoded images, returns JSON palette data