<![CDATA[The goal this project is to create a face-shape classification and hairstyle recommendation system by combining Support Vector Machine (SVM) and Random Forest (RF) algorithms with Histogram of Oriented Gradients (HOG) feature extraction. This study is motivated by the growing demand for individualized appearance support, as many users find it difficult to find haircuts that complement their face features. The method first preprocesses facial photos, uses HOG to extract key geometric and texture-based features, and then uses SVM and RF models to categorize the images. For training, validation, and testing, a dataset of five different face shapes is utilized. According to experimental results, the Random Forest model has an accuracy of about 89%, while the SVM model achieves an accuracy of about 95%. These findings suggest that SVM is better suited for managing high-dimensional feature spaces generated by HOG extraction. A recommendation system that offers hairstyle recommendations based on the anticipated face shape is then integrated with the trained model. The system is useful for real-time use since it can process pictures taken with the camera or uploaded from the gallery. Overall, this study shows that integrating HOG with SVM offers a dependable basis for creating customized hairdo recommendations as well as an efficient method for face-shape classification. ]]>
Copyrights © 2025
