Gender Classification from Facial Features Using Logistic Regression and K-Nearest Neighbors
Vijay Madaan, Neha Vaishnavi Sharma, Sripelli Jagadish, Haayder M. Abbas, Gaurav Sharma · 2025
This paper presents a gender classification model based on face features using logistic regression and K-Nearest Neighbors (KNN) classifier. The dataset contained long hair, forehead breadth, nose size, and lip thickness among other features. Development and evaluation of logistic regression and KNN models were done using a 60-40 split for both training and testing. Trained over 50 epochs, the models had a batch size of 32 and learning rates of 0.001 for Logistic Regression and 0.01 for KNN. With an accuracy of 95.37%, the Logistic Regression model performed; KNN exceeded with an accuracy of 96.32%. Regularizing Logistic Regression lowered its final loss relative to KNN's, hence assuring less overfitting. Combining machine learning classifiers with facial feature extraction for gender classification exhibits great efficacy in this work.