Open Access Open Access  Restricted Access Subscription Access

AI-Based Beauty Advisor with Privacy Guardian for Intelligent Skin Condition Analysis

Anirudha Sanjay Sahane, Mrs. Mahesh. D.Nirmal, Nikita Kishor Shende, Yash Devidas Nawale, Rajgauri Rajendra Parkhe

Abstract


With the advancement of Artificial Intelligence in the medical field and cosmetics industry, intelligent systems have been developed that can offer individualized skin analysis and suggestions. This study introduces the AI Skin Advisor, a mobile application that uses AI to identify prevalent skin conditions from facial images and offer skin care advice without compromising user privacy. The proposed system involves using a Convolutional Neural Network (CNN) with the MobileNetV2 architecture, trained through transfer learning for the classification of skin conditions like Acne, Redness, Dark Spots, Wrinkles, and Eye bags.

Built with the Flutter framework for cross-platform mobile development, the app also leverages TensorFlow Lite for on-device inference. Google ML Kit is used for face detection and validation for proper preprocessing before analyzing. It also utilizes Firebase Authentication and Realtime Database for user management, and the Google Gemini API for AI-powered skin care assistance via a chatbot interface.

The proposed system employs a privacy-focused approach, processing images on the user's own device and never storing or sharing personal images. To boost the performance of models on small datasets, data augmentation and transfer learning techniques are employed. The experimental results show that the system gives a quick and accurate prediction with real-time performance appropriate for the use of the Android device.

The proposed AI Skin Advisor system provides an easy-to-use, intelligent, and user-friendly approach to initial skin condition analysis and customized skin care recommendations. It is planned to enlarge the dataset, improve the accuracy of predictions, and add features for consulting with a dermatologist to provide better healthcare services in the future.


Full Text:

PDF

References


Deep Learning Goodfellow, I., Bengio, Y., & Courville, A. (2016). Deep Learning. MIT Press.

Dermatologist-level classification of skin cancer with deep neural networks Esteva, A., Kuprel, B., Novoa, R. A., et al. (2017). “Dermatologist-level classification of skin cancer with deep neural networks.” Nature, 542(7639), 115–118.

Classification of skin lesions using deep convolutional neural networks Han, S. S., Park, G. H., Lim, W., et al. (2018). “Classification of skin lesions using deep convolutional neural networks.” Journal of Digital Imaging.

Skin disease diagnosis using deep learning algorithms Li, X., Zhang, S., & Wang, J. (2019). “Skin disease diagnosis using deep learning algorithms.” IEEE Access.

AI in personalized skincare: Hybrid models for skin type recognition Wang, J., Chen, Y., & Liu, H. (2020). “AI in personalized skincare: Hybrid models for skin type recognition.” International Journal of Cosmetic Science.

Privacy-preserving AI for healthcare using federated learning Park, H., Kim, S., & Lee, J. (2021). “Privacy-preserving AI for healthcare using federated learning.” IEEE Transactions on Medical Imaging.

Flutter Documentation

Firebase Documentation

TensorFlow Lite Documentation

Google ML Kit Documentation


Refbacks

  • There are currently no refbacks.