Phish-Guard: A Visual Phishing Detection Approach Using Webpage Color Palette and Machine Learning

Phishing is a social engineering attack in which a hacker uses emotional tricks to trick people into revealing sensitive information or committing security mistakes. Machine-learning anti-phishing techniques are becoming increasingly popular. This study introduces Phish-Guard, a machine learning-powered framework which utilizes visual inspection to identify phishing websites. Phish-Guard detects phishing websites by combining color palettes and brand names encoded on web pages with machine learning. Phish-Guard uses five machine models to analyze its predictive performance of color palette features and popular brand names of web pages. The random forest model outperformed the other algorithms and detected phishing scams with a 98.27% recall rate and 99.02% classification accuracy. The Phish-Guard performance demonstrates the feasibility and usefulness of leveraging the webpage color palette features to identify phishing websites.