Face Recognition System

01 /Overview

Build a face detection and recognition web app that is light enough to deploy without heavy dependencies such as dlib.

02 /What I built

• Face detection with OpenCV’s DNN detector, falling back to Haar cascades
• Face alignment using eye detection
• Recognition with HOG features and an SVM classifier, trained on the LFW dataset
• Data augmentation: flips, rotations and brightness changes
• Model evaluation metrics shown in the app

03 /Outcome

A Streamlit app that detects, aligns and recognises faces from uploads or the camera, and reports accuracy, precision, recall and F1.

04 /What I learned

Classic computer vision still goes a long way: good detection and alignment mattered as much as the classifier.