Predictive Analysis of Skin Type and Personalized Acid Recommendations for Effective Skin Care Using ResNet-50

The largest organ in our body is the skin, made up of three layers: Epidermis (outermost), Dermis (middle), and Hypodermis (subcutaneous). Different skin tones range from fair to dusky. Different skin types include Normal (neither too oily nor dry), Oily (high oil production), Dry (low moisture levels), Combination (varying types in different areas), and Sensitive (prone to sun and virus damage). Understanding the skin type is crucial due to factors like Melanin production, Genetics, Climate, Hormones, Age, Health Conditions, Diet, Stress, Skin Care Routine, etc. Proper use of the right acid composition is essential for healthy skin maintenance. The current CNN method can only capture short-range dependencies and requires a huge number of labelled data for generalization due to its limitations in accessing long-range dependencies. The existing CNN used for skin type classification and acid recommendation encounters issues like imbalanced and limited datasets, overlapping features, subjective labelling, variability in images, and challenges in connecting skin types to appropriate assets. The suggested approach utilizes ResNet50, which offers several benefits compared to CNNs in predicting skin types. ResNet50 incorporates skip connections, also known as residual connections, which enable the network to achieve greater depth without experiencing a decrease in performance. ResNet50 enables more accurate feature extraction, leading to increased classification accuracy. Random Vertical Flip is utilized for image augmentation to flip images, boosting data diversity randomly. Following the system's recommended skincare routine will lead to a significant outcome, as our method has a higher accuracy compared to traditional methods.