How qver3nc1a/eigenfaces works
Eigenfaces is a face-recognition program: it prepares Olivetti face images, loads and splits them into training and validation sets, computes an eigenface representation, classifies validation faces by nearest neighbor, and reports accuracy. The PCA workflow is implemented in one module, while Jacobi eigen-decomposition is a separate numerical component. The graph follows the sampled source; documentation beyond the README was not inspected, so details of the user guide are not inferred.
Main entry points: pca.py PCA workflow · load_olivetti.py Dataset fetcher · dataset.py Split and preprocess · data Face image files
Components 12
Face data
- Dataset fetcher
data/load_olivetti.py - Face image files
data - Split and preprocess
src/dataset.py
Recognition workflow
- PCA workflow
src/pca.py - Mean centering
src/pca.py - Eigenface basis
src/pca.py - Face projection
src/pca.py - Nearest neighbor
src/pca.py - Accuracy report
src/pca.py
Numerical methods
- Jacobi eigensolver
src/jacobi.py
Other components
- User
- Olivetti dataset
Connections 11
- User → PCA workflow: runs
README.md - Dataset fetcher → Olivetti dataset: fetches
data/load_olivetti.py - Dataset fetcher → Face image files: writes images
data/load_olivetti.py - Split and preprocess → Face image files: reads images
src/dataset.py - PCA workflow → Split and preprocess: loads splits
src/pca.py - PCA workflow → Mean centering: centers training faces
src/pca.py - PCA workflow → Eigenface basis: builds basis
src/pca.py - PCA workflow → Jacobi eigensolver: decomposes covariance
src/pca.py - PCA workflow → Face projection: projects faces
src/pca.py - PCA workflow → Nearest neighbor: predicts labels
src/pca.py - PCA workflow → Accuracy report: computes accuracy
src/pca.py