DermaLens: Facial Skin Analysis and Personalized Skincare Recommendation System using Computer Vision Technique

DermaLens: Facial Skin Analysis and Personalized Skincare Recommendation System using Computer Vision Technique

Tousif Inamdar1, Shakila Sidavattam2

1Masters Student, Department of Computer Science, Abeda Inamdar Senior College, India

2Head of Department, Department of Computer Science, Abeda Inamdar Senior College, India

Abstract – Skin health is very important for a persons appearance and confidence. Recently, many people have been interested in understanding their skin issues and treating them well by following a skincare routine. Despite this, visiting a dermatologist can be time-consuming, expensive, and not easily available for everyone. Many existing skincare system provides general skincare advice or depend on complex deep learning models that need large datasets and powerful computing resources.

To address these issues, this research introduces a facial skin analysis and personalized skincare routine recommendation system using computer vision(OpenCV), basic machine learning, and OpenAI API. The System analyzes a users facial image to find common skin issues such as pimples, acne, dark spots, pigmentation, and dryness. It also finds out the users skin type and provide personalized skin care routine with home remedies. The system is built using React.js for the frontend, Flask, Node.js, and Express.js for the backend, OpenCV for image processing and feature extraction, and OpenAI API for skin care routing recommendation.

This paper describes the design, development, and implementation of the skin analysis Application, focusing on its architecture, technologies, and practical functionality. The developed solution provides an easily available, accessible, and affordable option for initial skin assessments. This application is available for users who want to know their skin condition and skin care routine without a visit to a dermatologist.

Keywords: Facial Skin Analysis, Computer Vision, OpenCV, Machine Learning, Skincare Recommendation, Skin Type Detection, Web Application, AI in Healthcare.

1. INTRODUCTION

   1. Problem Statement

      Problems related to the skin, such as acne, skin pigmentation, dark spots, patches of skin with varied tones, and dry skin, are commonly encountered by people of all age groups. These are affected by factors such as pollution, lifestyle, stress, climate, and inappropriate skin care practices. Although consultants with thorough knowledge of skin-related issues are available to provide correct diagnoses and remedies, visiting them periodically is not possible due

      to high charges, busy schedules, and unavailability, especially in the case of students or in semi-urban or rural regions.

      The available solutions for them may raise generalized recommendations without interpretatively studying their skin. Furthermore, most of the latest systems for analyzing their skin may depend completely on deep learning models like Convolutional Neural Networks. They need lots of resources. They cannot be used for lightweight systems. Therefore, a need may arise for developing a system that may interpretatively analyze their facial skin and suggest them appropriate skincare recommendations without depending on computationally intensive models.

   2. Significance

      The lack of an accessible digital platform for skin analysis limits people's ability to understand their skin condition in the early stages. Having an initial online solution for analyzing facial skin could help people take the appropriate skin care measures before it is too late. This system can also increase awareness about skin hygiene and encourage a person to take care of their skin.

      With the application of computer vision concepts along with basic machine learning, this system is able to minimize its dependency on deep learning while providing valuable outcomes. This makes it lightweight, faster, and appropriate for practical applications, especially in academic or small medical settings.

   3. Proposed Solution

   The skin analysis and skincare recommendation system allows users to upload a clear skin image through a web interface. The system analyzes the users skin image to detect common skin problems and the users skin type.

   The system suggests a personalized skincare routine and home remedies based on the skin analysis result. The web- based system is built using React.js for the frontend, Node.js, Flask for the backend, and OpenCV for image analysis.

   The goal of the system is to provide users with an accessible, cost-effective, and user-friendly platform for initial skin analysis.

2. LITERATURE REVIEW

   Recently, skin analysis and dermatological assessment have gained more attention because of the high demand for skincare and early detection of skin-related issues. Many studies have explored the use of image processing, computer vision, and artificial intelligence for analyzing skin issues.

   Zebari and Sallow proposed a face detection and recognition system based on OpenCV, with a focus on real-time detection and image pre-processing methods [1]. The study showed that OpenCVs Haar Cascade classifiers can accurately detect human facial regions in images and videos under varying conditions. This forms part of the face detection in DermaLens, where accurate segmentation of facial areas is important before performing skin analysis.

   Different methodologies related to skin detection have been analyzed by Nanni, Loreggia, and Lumini, who introduced a standardized approach to evaluate the effectiveness of skin detection techniques using multiple datasets[2]. The authors compared different color space models related to skin detection, highlighting the importance of using preprocessing techniques. This research supports the preprocessing techniques used in the DermaLens system, where accurate skin detection is important.

   Machine learning based skin image analysis tools have also been explored by Xiao et al., who developed such a system capable of analyzing skin images using a supervised learning algorithm[3]. Their approach demonstrated that supervised trained models can classify skin conditions based on visual feature extraction. This study confirms the suitability of the intelligence analysis mechanism in DermaLens with regard to skin-related conditions.

   Lin et al. investigated the application of computer vision to a skincare product recommendation system based on images[4]. The system recommends relevant cosmetic products based on facial features identified, demonstrating that computer vision could move a step further from only detection to offering recommendations. This concept ties with DermaLenss approach of pairing analysis with skincare products and home remedies recommendations.

   Feature extraction techniques of skin image detectors were further explored by Al-Zebari et al.,who then proposed scale-invariant feature extraction techniques for better detection accuracy [5]. Their work pointed out the need to extract features that are invariant to resolutions and scales, which align with DermaLens's need for scale-variant features.

   Similarly, Medjram et al. worked on the topic of improving skin detection accuracy by applying a color space transformation and a texture-based approach[6]. The results

   of this study also proved that by combining texture and color features, accuracy may be increased, which is also a part of the methodological framework adopted by DermaLens, where multiple characteristics are analyzed for skin issue detection.

   In summary, the reviewed literature proves that OpenCV- based face detection, standardized skin segmentation methods, machine learning classification, and feature extraction tecniques are effective in automated skin analysis systems. However, a majority of current solutions offer detection or recommendation functions independently. DermaLens unites these elements in one web-based platform that offers facial detection, skin issue analysis, skincare product recommendation, and home remedies all in one system.

3. METHODOLOGY

   1. Design of Research

      To develop a web application that analyzes facial skin, identifies issues, and suggests a skincare routine with home remedies and dermatological products. The focus of this research is on building a user-friendly web application using OpenCV and the OpenAI API, rather than complex deep learning models. Through requirement analysis, system design, development, testing, and evaluation. The goal is to ensure that the system is efficient, reliable, and capable of performing skin analysis accurately while keeping computational demands low. The goal of the design is to be used by individuals with less technical skills.

   2. Information Gathering

      • Secondary Data

        Secondary data was collected through an in-depth review of existing journals, research papers, and technical articles related to computer vision, skin analysis, and an AI-based skin-care web application. These studies provided awareness of common skin problems, image processing techniques, and limitations of the existing skin analysis systems. Literature related to non-deep learning based image analysis was also reviewed to justify the selection of lightweight and efficient techniques.

      • Technical Research

        In the technical research, different methods and tools of image processing were studied for the facial skin analysis web application. OpenCV has been used for image preprocessing and feature extraction due to its efficiency and applicability in computer vision systems. In the backend development of the web application, Node.js, Express.js, and Flask have been used. React.js has been used to develop the frontend of the web application, and an Open API key is used for recommending skin care routines.

   3. Architecture of the system

   The system is developed using the MERN stack, consisting of MongoDB, React.js, Express.js, Node.js, and OpenCV for image analysis.

4. DESIGN AND IMPLEMENTATION

1. System Architecture

   The DermaLens system follows a three-tier architecture and contains:

   Frontend(Client-Side): A React.js based responsive web interface that allows user to upload their image and show their skin condition with a skincare routine.

   Backend(Server-Side): The backend is developed using Node.js, Express.js, and Flask, which handles image processing requests and performs skin analysis using OpenCV.

   Database(MongoDB): Stores user information, image analysis results, and skincare routine data.

   • System workflow

     1. The authenticated user gets access to the web interface to upload their facial skin image.

     2. The facial image of the user is sent to the backend server for processing.

     3. Image preprocessing and skin feature extraction are done using OpenCV.

     4. The system analyses skin issues and finds the skin type.

     5. Based on the skin condition and type, the system provides a suitable skin care routine.

     6. The results are displayed to the user, which is easy to understand.

   • System Architecture Diagram

     Figure 1 System Architecture of DermaLens

2. Technologies Used

   The DermaLens system contains the following technologies

   :

   Table 1 Technology Stack for DermaLens

   Component	Technologies Used
   Frontend	React.js, HTML, CSS, MaterialUI, JavaScript
   Backend	ode.js, Express.js, and Flask
   Image Processing	OpenCV
   Database	MongoDB
   API Communication	REST API
   Authentication	JSON Web Token (JWT)

3. User Interface(UI) & Screenshots

   The DermaLense web application is designed to maintain the simplicity of its interface, ensuring smooth interaction with users who perform facial skin analysis. The interface is responsive and accessible on both desktop and mobile browsers. The design focuses on clarity, ease of navigation, and minimum effort by the user in uploading images and viewing the results of the analysis.

   1. User Interface Overview

      The system offers the following main interface screens:

      • Homepage: Provides an overview of the DermLens web application, including features, a signup/login option, and a start analysis button to navigate to start analysis.

      • Signup Page: New users can sign up to DemaLense.

      • Login Page: Facilitates the login process for registered users who can then use the application for skin analysis.

      • Image Upload Page: Allows users to upload a clear image of their face for analysis.

      • Processing Screen: It displays a loading interface while the system performs image preprocessing, feature extraction, and skin analysis.

      • Result Page: Displays the identified skin type, recognized skin issues, and provides a personalized skincare routine in an organized manner.

   2. UI Screenshots

The following figures illustrate the key UI screens of the DermaLens web application:-

Figure2: Homepage of DermaLens

Figure 3: Signup/Register page

Figure 4: Login Page

Figure 5a: Image Upload Interface

Figure 6b: Skin Analysis Result

Figure 7c: Skincare Recommendation Screen

Figure 8d: Natural Remedies and Product

6. DISCUSSION

1. Strengths of the System

   • Lightweight Implementation: Computer vision is used to develop this system, which is simpler than complex deep learning techniques.

   • Cost-Effective Skin Evaluation: The user can gain some idea about their skin condition without requiring regular visits to dermatologists, which helps them save time and money.

   • User-Friendly Interface: The simple and attractive interface enables anyone to upload images, regardless of their technical expertise.

   • Personalized Recommendations: As compared to other skincare applications, DermaLense gives a skincare routine based on the users skin type and problem.

   • Real-Time Processing: Preprocessing and analysis of images are carried out effectively, and results are available within a few seconds.

   • Educational Awareness: The system provides greater awareness to users regarding skin care and promotes preventive skin care practices.

2. Limitations of the System

   • Dependence on Image Quality: The accuracy of skin analysis, however, would depend on the clarity, light, and position of the uploaded image of the face.

   • No Deep Learning Model: Although this system is lightweight, detection accuracy for complicated conditions might still be affected by the lack of CNN- based models.

   • Limited Skin Issue Coverage: The system deals only with average problems like acne, dryness, and pigmentation. There is no inclusion of rare skin.

   • No Clinical Validation: The system is intended to carry out preliminary analysis, not to be used as a substiute for professional medical diagnosis.

3. Future Scope

   • Deep Learning Integration: Deep learning models like Convolutional Neural Networks (CNNs) can be

     incorporated to improve the accuracy of detection of complex skin conditions like severe acne, pigmentation, and wrinkles.

   • Dermatologist Consultation Feature: The online dermatologist consultation can be included to allow users to get professional advice.

   • Mobile Application Development: A mobile application can be developed to facilitate real-time skin analysis using smartphone cameras.

1. CONCLUSION

   Facial skin health is a key determinant of personal grooming and self-confidence, but seeking advice from a dermatologist may not always be economically viable and convenient. Many existing intelligent facial skin analysis solutions have relied on computationally intensive deep learning approaches, making them hard to integrate into web contexts. In an effort to bridge this gap, an intelligent facial skin analysis and skincare advice system known as DermaLens, which utilizes computer vision and image processing techniques.

   This has helped the system remain an efficient, fast, and practical solution. It is important to note that the system works efficiently depending on the image quality. The solution is not intended for a medical diagnosis, but works well as an effective system for the initial skin evaluation and awareness. With the improvements to come in the form of an advanced dermat Detection technique, Mobile Application development, and the expansion of the current application of features, DermaLens has the possibility of evolving.

2. REFERENCES

1. 1.R. Zebari and A. Bibo Sallow, Face Detection and Recognition Using OpenCV, Journal of Soft Computing and Data Mining, Vol. 2, No. 2, Oct. 2021.

2. 2.L. Nanni, A. Loreggia, A. Lumini, A Standardized Approach for Skin Detection: Analysis of the Literature and Case Studies, Journal of Imaging, Vol. 9, No. 2, 2023.

3. 3. P. Xiao et al., The Development of a Skin Image Analysis Tool by Using Machine Learning Algorithms, Cosmetics, vol. 7, no. 3, 67, 2020.

4. 4.Facial Skincare Products Recommendation with Computer Vision Technologies, Electronics, MDPI, 11(1), 143, 2024.

5. 5. T.Y. Lin, H. T. Chan, C. H. Hsia, and C. F. Lai Facial Skincare Products Recommendation with Computer Vision Technologies, Electronics, vol. 11, no. 1, p. 143, Jan. 2022.

6. A. Al-Zebari, S. W. Kareem, and S. H. I., Scale-Invariant Feature Extraction for Skin Image Detection, Int. J. Intell. Syst. Appl. Eng., vol. 12, no. 21s, 2024.

7. S. Medjram, M. C. Babahenin, and Y. Benali, Improved skin detection using Colour Space and Texture, Int. J. Intell. Syst. Appl. Eng. , vol. 12, no. 21s, 2024.