Efficient Net-based Expert System for Personalized Facial Skincare Recommendations

In recent years, recommender systems have gained a lot of traction in both the commercial world and the academic research community. The term “recommendation system for facial skin care” refers to an automated system that gives consumers advice on which facial skin care products to use based on their specific skin type, skin issues, and preferences. Product recommendation systems analyze user data using a variety of methods, including collaborative filtering, content-based filtering, and machine learning algorithms. Better skincare results may be achieved through the usage of a suggestion system in which consumers receive advice related to their own skin issues. The problem with current models is that they don’t adequately address users’ issues and concerns while being based on collaborative or content-based filtering. In order to create an effective face skincare model, the authors of this study suggest a hybrid approach that integrates several existing methods (including KNN, CNN, Transfer Learning of Efficient Net BO, and a content-based filtering recommendation engine). The user’s skin tone, skin type, and acne severity are all inputs into the algorithm, which then outputs a recommendation for the most suitable product. The results of a comparison between the suggested model and other models prove its superiority. Using a hybrid model that takes into account a number of different methodologies and user inputs, the article aspires to deliver a more all-encompassing and individualized solution for facial skincare recommendations. The proposed model has the potential to enhance the precision and efficiency of current skincare recommendation systems. The implementation of EfficientNet B0 improves the model’s accuracy, which now has a validation accuracy of 80% and a training accuracy of 87.10%.