Olivetti Faces (Classification with ai model)
تفاصيل العمل
Dataset: Olivetti Faces — 400 grayscale face images for 40 persons (10 images each), augmented to 2000 samples. What we did: We loaded and visualized the dataset, applied data augmentation (rotation, flip, brightness) to expand the data 5x, then standardized features and reduced dimensionality with PCA (150 components, 92% variance retained). We trained 5 classifiers — KNN, Random Forest, XGBoost, Decision Tree, and Logistic Regression — evaluated them with classification reports and confusion matrices, then implemented Logistic Regression from scratch using softmax + gradient descent and compared it against sklearn's version, achieving 98.33% vs 98.67% accuracy.
مهارات العمل
