EfficientV2 for Real-Time Skin Type Detection with Enhanced Image Processing

In this modern world of fast-paced life, no one really do not care about skin care. Skincare is one of the things left behind. Skin texture analysis is really important for several purposes such as dermatology, skin texture modeling, face recognition, etc. Skin type recognition also helps us in diagnosing several skin diseases that can be both hereditary or acquired. There are several Image processing techniques that can help us classify the skin texture, but mostly require heavy computation and are slow in real-time. In this proposed research paper, discussion is done on a deep learning model EfficientV2 which focuses on designing neural network structures that require fewer parameters and computations compared to traditional architectures. Further to help in increasing the accuracy of the result that is obtained, the image is being processed with methods like Contrast Limited Adaptive Histogram Equalization (Contrast Limited Adaptive Histogram Equalization) which is an image processing algorithm that helps by enhancing the contrast while keeping the noise of the image as low as possible. Also, augmentation is being done on the image to increase the accuracy of the proposed method. The accuracy that is being obtained from using the following methodology is 98.04% and also the model size obtained is 22.59 megabytes (MB). This proposed methodology of real-time skin detection method is not fast but also has a high accuracy percentage.