Eigenfaces: Facial Feature Extraction & PCA Visualizer

[READY: CANONICAL SET]

DRAG PUCK TO MORPH LATENT SPACE 60 FPS

🎙️ OVERVIEW AUDIO GUIDE Introduction

00:00 / 02:40

DATASET FACES (M) 8

TOTAL PIXELS (D) 4,096

EXPLAINED VAR. 91.4%

RECON ERROR (RMSE) 0.042

1. Image Acquisition & Library

Training Gallery (Click to Target for Projection):

Canonical dataset

2. PCA Parameters & View Mode

Reconstruction Visualization Mode

Image Resolution (W × H) 64x64

Retained Components (K) 4 / 7

Diagnostics & Subspace View

3. Eigenface Weight Sliders (ωₖ)

Standardized by standard deviation σₖ = √(λₖ / M). Drag sliders or grab the 2D Latent Puck on the canvas.