Enhanced Image Classification of African Women: A Synergy of PCA and Random Forest Methods

This study introduces a novel approach to classifying facial images of African women, addressing a gap in comprehensive classifiers for African hairstyles in the digital era. The paper employs semi -supervised learning techniques to tackle this challenge. The methodology involves pre-processing facial images through Principal Component Analysis (PCA) to reduce dimensionality, thereby improving the efficiency of model training. Subsequently, K-Means clustering is utilized to systematically categorize the images into distinct groups, forming the basis for further classification. The classification process employs the Random Forest algorithm, well-known for its robustness and accuracy in image classification tasks. Modeling and evaluation leverage FigarolK and Flickr datasets, implementing image augmentation to address the constraints of limited dataset size. The Random Forest model achieves notable accuracy, classifying 89% of test images with an 11 % misclassification rate. These results highlight the approach's potential contribution to facial image classification, particularly for African women, amid digital transformation.